JP5520505B2 - Scoring apparatus, scoring method, and program - Google Patents

Description

  The present invention relates to a scoring device for scoring a character string input by handwriting.

  Conventionally, when the stroke of the input character is displayed on the display during input as a path connecting a straight line and a quadratic Bezier curve, the starting point / end point / folding point / curve of the stroke is determined. There has been a handwritten character input system that displays a feature point such as a graphic mark (see, for example, Patent Document 1).

JP 2008-108177 A (first page, paragraph [0059], etc.)

  However, in the conventional handwritten character input system, automatic scoring and manual scoring were not effectively combined, so scoring with high accuracy was not possible.

  Moreover, in the conventional handwritten character input system, there is a problem that the processing of paying attention to the structural points that are the end points, the branch points, or the intersections constituting the character skeleton is not performed, and the speed of the character correctness determination processing is slow. there were.

  Moreover, in the conventional handwritten character input system, the structure point and the feature point are distinguished, and different algorithms are applied to each to perform the character correctness determination process. There was a problem.

  The scoring device according to the first aspect of the present invention can store a correct character information storage unit that can store one or more correct character information indicating a correct character string, and a character image that is an image including a character string input by handwriting. A character image storage unit; a determination unit that compares the character image with the correct character information and obtains a determination result indicating whether the handwritten character string is correct or incorrect; and a determination to output the determination result A result output unit; a reception unit that receives a user determination result indicating whether the character string input by handwriting is correct or incorrect after the determination result output unit outputs the determination result; and the reception It is a scoring device comprising a determination result storage unit that stores user determination results received by the unit.

  With such a configuration, it is possible to efficiently score correct and incorrect character strings.

  In the scoring device of the second aspect of the invention, in contrast to the first aspect of the invention, the accepting unit accepts the user judgment result and the judgment reason, and the judgment result accumulating unit is the user judgment result and the judgment reason accepted by the accepting unit Is a scoring device that accumulates

  With this configuration, it is possible to accumulate the determination results and determination reasons of the user and use the determination reasons later.

  The scoring device according to the third aspect of the present invention has correct character information including a correct character string and determination information which is information for correct / incorrect determination for either the first or second invention, The scoring device further includes a determination information update unit that updates the determination information when the determination result in the determination unit is different from the user determination result received by the reception unit.

  With this configuration, it is possible to accurately determine whether a character string is correct.

  In the scoring device according to the fourth aspect of the present invention, as compared with the first to third aspects, the determination information has one or more feature amounts that characterize the correct character, and the determination unit has one or more feature amounts. This is a scoring device that determines whether or not the character included in the character image is incorrect using the determination information including

  With this configuration, it is possible to accurately determine whether a character string is correct.

  In the scoring device according to the fifth aspect of the present invention, the correct character information is information on a structure point that is an end point, a branch point, or an intersection that constitutes the character skeleton. The determination unit has structure point information, and the determination unit detects one or more structure points for each character included in the character image stored in the character image storage unit, and determines the structure point information that is information on the structure point. One or more point information acquisition means to be acquired, the structure point information included in correct character information corresponding to each character included in the character image, and the structure point information acquired by the point information acquisition means are compared, and The scoring device includes a determination unit that determines whether each character included is an incorrect answer.

  With this configuration, it is possible to determine whether a character string is correct or incorrect.

  Further, the scoring device of the sixth aspect of the invention is that, with respect to the fifth aspect, the correct character information is structure point information that is information of a structure point that is an end point, a branch point, or an intersection that constitutes the skeleton of the character, A point on the edge that is a line connecting two structural points, and feature point information that is feature point information that is a point having a degree of bending larger than a threshold value. For each character included in the character image stored in the character image, the structure point information acquisition means for acquiring one or more structure point information, and for each character included in the character image stored in the character image storage unit And a feature point information acquisition unit that acquires one or more feature point information, and the determination unit uses one or more structures acquired by the structure point information acquisition unit according to different uses of the structure point information and the feature point information. 1 or more acquired by the point information and feature point information acquisition means And the feature point information is compared with one or more structural point information and one or more feature point information included in the correct character information, the character included in the character image is a scoring device for determining solved fraud.

  With this configuration, it is possible to accurately determine whether a character string is correct.

  Further, in the scoring device of the seventh invention, in contrast to the sixth invention, the judging means has one or more structure point information acquired by the structure point information acquiring means and one or more structures included in the correct character information. The structure determination means for comparing the point information and determining whether the character included in the character image is incorrect or not, and the structure point for the character that the structure determination means determines that the character included in the character image is not incorrect. One or more structure point information acquired by the information acquisition means, one or more feature point information acquired by the feature point information acquisition means, and one or more structure point information and one or more feature point information included in the correct character information. It is a scoring device comprising a feature determination means for comparing and determining whether or not a character is an incorrect answer.

  With this configuration, it is possible to determine whether a character string is correct or not at high speed and with high accuracy.

  Further, in the scoring device of the eighth invention, in contrast to the seventh invention, the feature judging means includes at least one structural point or feature point included in one character having a connected structural point or feature point. Included in correct character information, group dividing means for dividing into groups, one or more structure point information obtained by the structure point information obtaining means and one or more feature point information obtained by the feature point information obtaining means for each group And a group feature judging means for comparing one or more structural point information and one or more feature point information to judge whether or not the character is an incorrect answer.

  With such a configuration, it is possible to determine the correctness of the character string with higher accuracy.

  The scoring device according to the ninth aspect of the invention relates to the eighth aspect of the invention, in which the feature determination means is a feature quantity acquisition means for acquiring one or more feature quantities indicating a balance between groups of handwritten characters. The scoring device further comprises balance determination means for determining whether or not the character is an incorrect answer using one or more feature quantities acquired by the feature quantity acquisition means.

  With such a configuration, it is possible to determine the correctness of the character string with higher accuracy.

  The scoring device according to the tenth aspect of the invention is the eighth or ninth aspect of the invention, wherein the group feature judging means is one or more pieces of structure point information and features acquired by the structure point information acquiring means for each group. It is a point corresponding to each point indicated by one or more feature point information acquired by the point information acquisition means, and indicates either one or more structure point information or one or more feature point information included in the correct character information. Corresponding point information that is point information is acquired, and the distance between each point indicated by each of the one or more structural point information and one or more feature point information and each point indicated by each corresponding point information is calculated, This is a scoring device that calculates the sum of absolute values and determines that the character is incorrect when the sum of absolute values of distances is greater than or equal to or greater than a threshold value.

  With such a configuration, it is possible to determine the correctness of the character string with higher accuracy.

  The scoring device according to the eleventh aspect of the present invention is the ninth or tenth aspect of the present invention, wherein the feature amount acquisition means occupies an area occupied by one or more groups of characters input by handwriting. The ratio is calculated as a feature amount, and the balance determination means acquires determination information including the ratio of the area occupied by a character portion composed of one or more groups of correct characters, and the feature calculated by the feature amount acquisition means It is a scoring device that determines whether or not a character is an incorrect answer from an amount and determination information.

  With such a configuration, it is possible to determine the correctness of the character string with higher accuracy.

  The scoring device according to the twelfth aspect of the invention relates to the ninth or tenth invention, wherein the feature amount acquisition means calculates the distance between each group of handwritten characters as a feature amount, and determines the balance. The means is a scoring device that acquires determination information including a distance between groups of correct characters and determines whether or not the character is an incorrect answer from the feature amount calculated by the feature amount acquisition unit and the determination information.

  With such a configuration, it is possible to determine the correctness of the character string with higher accuracy.

  Further, the scoring device of the thirteenth invention is a scoring device in which the character is a Chinese character as compared with any of the first to twelfth inventions.

  With this configuration, the correctness / incorrectness of kanji can be determined with high accuracy.

  According to the scoring device according to the present invention, correct and incorrect scoring of character strings can be performed.

Block diagram of scoring device in Embodiment 1 Conceptual diagram explaining the graph data generation process Conceptual diagram explaining the structure comparison process Conceptual diagram explaining the feature comparison process Conceptual diagram explaining the feature comparison process Flow chart explaining operation of the scoring device Flow chart explaining the operation of the automatic scoring process A flowchart for explaining the operation of the tie information acquisition process A flowchart for explaining the operation of the determination process A flowchart for explaining the operation of the structure comparison process A flowchart for explaining the operation of the shape comparison process A flowchart for explaining the operation of the feature amount comparison process Flow chart for explaining the operation of the manual scoring process Flowchart explaining operation of post-processing The figure which shows the example of the same correct character information management table The figure which shows the example of the same correct character information management table The figure which shows the output example of the same judgment result The figure which shows the example of an input screen of the same user judgment result The same judgment result output screen and the figure which shows the feature quantity correction processing Diagram showing the simulation screen The same judgment result output screen and the figure which shows the feature quantity correction processing Diagram showing the simulation screen Overview of the computer system Block diagram of the computer system

  Hereinafter, embodiments of a scoring device and the like will be described with reference to the drawings. In addition, since the component which attached | subjected the same code | symbol in embodiment performs the same operation | movement, description may be abbreviate | omitted again.

  (Embodiment 1)

  In the present embodiment, a character string scoring device using both automatic scoring and manual scoring will be described.

  Further, in this embodiment, a character string scoring device that can input a determination reason when manual scoring is incorrect will be described.

  Further, in the present embodiment, a scoring device that can improve the automatic scoring algorithm using the result of manual scoring will be described.

  In the present embodiment, a scoring device that acquires a structure point from an input handwritten character (image), compares it with a correct character using the structure point, and outputs a comparison result will be described.

  Further, in the present embodiment, a scoring device that performs structure comparison using structure points of handwritten characters and then performs feature comparison using structure points and feature points will be described.

  FIG. 1 is a block diagram of a scoring device in the present embodiment.

  The scoring device 1 includes a character image reading unit 10, a correct character information storage unit 11, a character image storage unit 12, a determination result storage unit 13, a determination unit 14, a determination result output unit 15, a reception unit 16, a determination result storage unit 17, A judgment information update unit 18 and a comparison output unit 19 are provided.

  The determination unit 14 includes a character division unit 141, a point information acquisition unit 142, and a determination unit 143.

  The point information acquisition unit 142 includes a structure point information acquisition unit 1421 and a feature point information acquisition unit 1422.

  The determination unit 143 includes a structure determination unit 1431 and a feature determination unit 1432.

  The feature determination unit 1432 includes a group division unit 14321, a group feature determination unit 14322, a feature amount acquisition unit 14323, and a balance determination unit 14324.

  The character image reading unit 10 reads one or more characters input by handwriting. The character image reading unit 10 forms an image including the character from one or more characters written on a medium such as paper. The character image reading unit 10 is, for example, a scanner or a digital camera.

  The correct character information storage unit 11 can store one or more correct character information. Correct character information is information indicating a correct character string. A character string is one or more characters. The character is preferably a Chinese character, but is not necessarily a Chinese character. In the present embodiment, the characters are not limited to general characters (such as hiragana, katakana, English characters, Korean, Arabic characters, etc.), and may be pictographs, symbols, symbols, and the like. Further, in the present embodiment, the character may be a figure or the like from which a structure point or / and a feature point can be extracted.

  Information of characters (hereinafter referred to as “correct characters” as appropriate) constituting the character string indicated by the correct character information may include structure point information, feature point information, and connection information as described later, or structure point information. It may be feature point information, a structure point only, or a character code. The correct character string is a character string to be compared with a character string input by handwriting. Note that characters constituting the character string input by handwriting are appropriately referred to as “input characters”.

  The structure point information is information indicating the structure points constituting the character skeleton. A structure point is an end point, a branch point, or an intersection that constitutes the skeleton of a character. The structure point information usually has position information (x, y) indicating the position of the point. A flag indicating a structure point may be added to the correct character information so that the structure point can be identified. The correct character information preferably includes one or more feature point information. The feature point information is feature point information. A feature point is a point on an edge, which is a point having a degree of bending larger than a threshold value. An edge is a line connecting two structural points. The correct character information preferably includes one or more pieces of determination information indicating character characteristics. The determination information is information indicating a feature amount or a feature amount range. The feature amount is, for example, the ratio of the area occupied by a group, which will be described later, the ratio of the area occupied by a character portion that is one or more groups, the distance between groups, the length of character splash, and a specific line (drawing). Any value can be used as long as it indicates the character of the character, such as the length of. A group is a set of connected structure points or feature points that constitute a character skeleton. The group may be a diagram composed of a set of connected structural points or feature points constituting a character skeleton. More specifically, it is preferable that the correct character information includes, for example, node information that is structure point information or feature point information, and connection information that indicates connection of structure points or feature points.

  The correct character information storage unit 11 is preferably a non-volatile recording medium, but can also be realized by a volatile recording medium.

  The process in which correct character information is memorize | stored in the correct character information storage part 11 is not ask | required. For example, correct character information may be stored in the correct character information storage unit 11 via a recording medium, and correct character information transmitted via a communication line or the like is stored in the correct character information storage unit 11. The correct character information input via the input device may be stored in the correct character information storage unit 11.

  The character image storage unit 12 can store a character image (referred to as “input character image” as appropriate) that is an image including a character string input by handwriting. The character image storage unit 12 normally stores a character image acquired by the character image reading unit 10. The character image storage unit 12 is preferably a non-volatile recording medium, but can also be realized by a volatile recording medium. The process of storing the character image in the character image storage unit 12 does not matter. For example, a character image may be stored in the character image storage unit 12 via a recording medium, and a character image transmitted via a communication line or the like is stored in the character image storage unit 12. Alternatively, the character image input via the input device may be stored in the character image storage unit 12.

  The determination result storage unit 13 can store the determination result. The determination result is a result of determination performed by the determination unit 14. The determination result storage unit 13 may also store a user determination result input by the user. The determination result storage unit 13 may also store information for identifying a character string to be determined (for example, an answer sheet identifier for identifying an answer sheet, an identifier for identifying a subject, etc.). The determination result storage unit 13 is preferably a non-volatile recording medium, but can also be realized by a volatile recording medium. The process of storing the determination result in the determination result storage unit 13 does not matter. For example, the determination result may be stored in the determination result storage unit 13 via a recording medium, and the determination result transmitted via a communication line or the like is stored in the determination result storage unit 13. Alternatively, the determination result input via the input device may be stored in the determination result storage unit 13.

  The determination unit 14 compares the character image (input character image) with the correct character information, and acquires a determination result indicating whether the character string input by handwriting is correct or incorrect. For example, the determination unit 14 performs character recognition processing on a character image and acquires a character code string. Then, the determination unit 14 compares the acquired character code string and correct character information (in this case, “one or more character codes”). If both companies match, the determination unit 14 determines that the answer is correct. Judge. Further, it is preferable that the determination unit 14 performs a determination process using a character division unit 141, a point information acquisition unit 142, and a determination unit 143, which will be described later. For example, the determination unit 14 preferably uses the determination information including one or more feature amounts to determine whether or not the character included in the character image is an incorrect answer.

  Here, the comparison does not require a direct comparison between the character image and the correct character information. For example, you may compare with correct character information using the structure point information and feature point information extracted from the character image. Further, as described above, the determination unit 14 may compare the character code string obtained by character recognition processing of the character image with the correct character information.

  The determination unit 14 can usually be realized by an MPU, a memory, or the like. The processing procedure of the determination unit 14 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The character dividing unit 141 cuts out each character from two or more characters included in the character image stored in the character image storage unit 12. Since the character dividing means 141 is a known technique used in character recognition processing or the like, detailed description thereof is omitted. The character dividing unit 141 can usually be realized by an MPU, a memory, or the like. The processing procedure of the character dividing unit 141 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The point information acquisition unit 142 acquires one or more structure point information and one or more feature point information for each character included in the character image stored in the character image storage unit 12. Further, the point information acquisition unit 142 may acquire only one or more structural point information for each character included in the character image stored in the character image storage unit 12. Each character included in the character image stored in the character image storage unit 12 is each character extracted by the character dividing unit 141. More specifically, the point information acquisition unit 142 graphs each character included in the character image, detects a node and an edge (line segment), and detects an end point, a branch point, or an intersection that is a specific node. This particular node is a structure point. Further, the point information acquisition unit 142 detects a feature point that is a point on the edge, which is a line connecting two structural points, and is a point having a degree of bending larger than a threshold value. The point information acquisition unit 142 stores a threshold value in advance. Further, the degree of bending is, for example, an angle formed by two straight lines. Further, the degree of bending may be a differential value (slope) of each point on the edge. Here, the node is a structure point or a feature point.

  The point information acquisition unit 142 can be usually realized by an MPU, a memory, or the like. The processing procedure of the point information acquisition unit 142 is usually realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The structure point information acquisition unit 1421 acquires one or more structure point information for each character included in the character image stored in the character image storage unit 12. More specifically, the structure point information acquisition unit 1421 operates as shown in FIG. 2, for example. First, the structure point information acquisition unit 1421 performs thinning processing on the handwritten character image. In the thinning process, the thickness is removed from the handwritten character image (a), converted to a line drawing image, and then converted from the line drawing image (raster) to line drawing data (vector (b)). Such conversion is called raster vector conversion. Next, the structure point information acquisition unit 1421 performs graph data creation processing. That is, the structure point information acquisition unit 1421 extracts a structure point from the line drawing data (vector). The structure point is an end point, branch point, or intersection of line drawing data.

  The structure point information acquisition unit 1421 can be usually realized by an MPU, a memory, or the like. The processing procedure of the structure point information acquisition unit 1421 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The feature point information acquisition unit 1422 acquires one or more feature point information for each character included in the character image stored in the character image storage unit 12. That is, the feature point information acquisition unit 1422 extracts feature points from the line drawing data (vector) in FIG. The process of the structure point information acquisition unit 1421 and the process of the feature point information acquisition unit 1422 constitute graph data (FIG. 2C) having nodes (structure points or feature points) and edges. Further, the point information acquisition unit 142 normalizes the non-normalized graph data (FIG. 2C), and obtains normalized graph data (FIG. 2D). Here, normalization is a process of storing node position information in a rectangular area of (0, 0) (0, 1) (1, 1) (1,0) in the XY coordinate system.

  The feature point information acquisition unit 1422 can be usually realized by an MPU, a memory, or the like. The processing procedure of the feature point information acquisition unit 1422 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The determination unit 143 compares the structure point information included in the correct character information corresponding to each character included in the character image with the structure point information acquired by the point information acquisition unit 142, and each character included in the character image is Determine whether the answer is incorrect. If the determination unit 143 determines from the two structure point information that the structures of the two characters do not match, the determination unit 143 determines that the two characters are different.

  The determination unit 143 uses one or more pieces of structure point information acquired by the structure point information acquisition unit 1421 and one or more pieces of feature point information acquired by the feature point information acquisition unit 1422 according to different uses of the structure point information and the feature point information. The one or more structure point information and the one or more feature point information included in the correct character information are compared, and it is determined whether or not the character included in the character image is an incorrect answer. Here, “different usage” means, for example, determining whether the structure matches or does not match using the structure point information, and determines whether the shape matches or does not match using the structure point information and the feature point information. Further, for example, “different usage” means determining whether the structure matches or does not match using the structure point information, and determines whether the character bending degree matches or does not match using the feature point information.

  The determination unit 143 preferably determines whether or not the character included in the character image is an incorrect answer based on one or more feature amounts received by the receiving unit 16. Here, reception means reception of information input from an input device such as a keyboard, mouse, touch panel, reception of information transmitted via a wired or wireless communication line, recording on an optical disk, magnetic disk, semiconductor memory, or the like. It is a concept including reception of information read from a medium. More specifically, the determining unit 143 preferably determines whether each character included in the character image is an incorrect answer by the structure determining unit 1431 and the feature determining unit 1432.

  The determination unit 143 can be usually realized by an MPU, a memory, or the like. The processing procedure of the determination unit 143 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The structure determination unit 1431 compares one or more pieces of structure point information acquired by the structure point information acquisition unit 1421 with one or more pieces of structure point information included in the correct character information, and determines whether or not the character included in the character image is an incorrect answer. Determine whether. The structure determining unit 1431 compares the structures of two characters and determines whether or not they are different characters. The structure determining unit 1431 preferably compares the structure of two characters using the connection information of the structure points to determine whether the characters are different. More specifically, the structure determination unit 1431 determines whether or not the character included in the character image is an incorrect answer by a method as shown in FIG. That is, the structure determination unit 1431 acquires the structure point information of the input character (here, connection information of each structure point) and the structure point information of the correct character (here, connection information of each structure point). Then, the structure determining unit 1431 determines that the two characters are different when the connection information (graph structure) included in the two structure point information does not match. Further, the structure determination unit 1431 may compare the position information of the structure point of the input character with the position information of the structure point of the corresponding correct character and calculate the difference. Then, the difference between the shapes of the two characters may be determined using the set of differences (for example, a value obtained by adding all the absolute values of the differences). That is, for example, the structure determination unit 1431 may determine that the two characters are different if the value obtained by adding all the absolute values of the differences is a predetermined value or more.

  The structure determination unit 1431 can be usually realized by an MPU, a memory, or the like. The processing procedure of the structure determination unit 1431 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The feature determination means 1432 has one or more structure point information and feature point information acquisition means acquired by the structure point information acquisition means 1421 for the character that the structure determination means 1431 determines that the character included in the character image is not incorrect. The one or more feature point information acquired by 1422 is compared with one or more structure point information and one or more feature point information included in the correct character information to determine whether or not the character is an incorrect answer. Note that the feature determination unit 1432 mainly compares two character shapes to determine whether or not the character included in the character image is an incorrect answer. Specifically, it is preferable that the feature determination unit 1432 determines whether the character is an incorrect answer by the group division unit 14321, the group feature determination unit 14322, the feature amount acquisition unit 14323, and the balance determination unit 14324. .

  The feature determination unit 1432 can be realized typically as an MPU, a memory, or the like. The processing procedure of the feature determination unit 1432 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The group dividing unit 14321 divides one character into one or more groups. That is, the group dividing unit 14321 divides the structure points and feature points included in one character into one or more groups having connected structure points or feature points. That is, the group dividing unit 14321 divides hiragana “i” into two groups, for example. The group dividing unit 14321 divides the Chinese character “字” into six groups, for example. As shown in FIG. 4, the group dividing unit 14321 determines that the connected node group is one group. The character “Po” is divided into three groups.

  The group dividing unit 14321 can be usually realized by an MPU, a memory, or the like. The processing procedure of the group dividing unit 14321 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The group feature determination unit 14322 includes, for each group, one or more pieces of structure point information acquired by the structure point information acquisition unit 1421 and one or more feature point information acquired by the feature point information acquisition unit 1422 and correct character information. The one or more structure point information and one or more feature point information are compared to determine whether the character is incorrect.

  The group feature determination unit 14322 corresponds to each point indicated by one or more structure point information acquired by the structure point information acquisition unit 1421 and one or more feature point information acquired by the feature point information acquisition unit 1422 for each group. Corresponding point information which is a point and is information of a point indicated by one or more structural point information or one or more feature point information included in the correct character information is acquired. Then, the group feature determination unit 14322 has a position of a point corresponding to each point indicated by the one or more structure point information acquired by the structure point information acquisition unit 1421 and one or more feature point information acquired by the feature point information acquisition unit 1422. The information is used to determine whether the character is incorrect.

For example, as shown in FIG. 5, the group feature determination unit 14322 determines a corresponding node for each group (in FIG. 5, the first group of “Po”), and calculates the distance between the nodes. . Then, the evaluation values of the two groups are calculated using the set of distances. If the evaluation value is greater than a predetermined value (including the above), the group feature determination unit 14322 determines that the two characters are different. Specifically, for example, as illustrated in FIG. 5, it is assumed that the group of the first stroke of the input character “PO” has nodes P1, P2, P3, and P4. Also, it is assumed that the group of the first stroke of the correct character “Po” has nodes Q1, Q2, Q3, Q4, and Q5. Then, the group feature determination unit 14322 determines that the node that is the other node and the closest distance between the groups of two characters is the corresponding node. In FIG. 5, the group feature determination unit 14322 determines that Q1, Q2, Q3, Q4, and Q5 correspond to P1, P2, P2, P3, and P4, respectively. Next, the group feature determination unit 14322 calculates the distances of the corresponding two points “(P1, Q1) (P2, Q2) (P2, Q3) (P3, Q4) (P4, Q5)”. Then, the group feature determination unit 14322 substitutes the distance into Equation 1 to calculate an evaluation value. Next, if the evaluation value is equal to or greater than the threshold value, the group feature determination unit 14322 determines that the input character is incorrect.

  In Equation 1, the position information of the point Pi is (pxi, pyi), and the position information of the point Qj is (qxj, qyj).

  The group feature determination unit 14322 can be usually realized by an MPU, a memory, or the like. The processing procedure of the group feature determination unit 14322 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The feature amount acquisition unit 14323 acquires one or more feature amounts indicating the balance of characters input by handwriting. Character balance is usually the balance between groups. In addition, the balance between groups includes the balance between one or more groups (character portions).

  For example, the feature amount acquisition unit 14323 calculates, for each group, the area of the minimum rectangle including each group from the position information of the nodes included in the group constituting each character. Then, the feature amount acquisition unit 14323 calculates the area of the minimum rectangle including the character from the position information of the node included in the character. Then, the feature amount acquisition unit 14323 calculates the ratio of the area of each group to the entire character by using the rectangular area of each group and the rectangular area of the character. In such a case, the feature amount is the ratio of the area of each group to the entire character.

  In addition, the feature amount acquisition unit 14323 calculates, as the feature amount, a ratio of the area occupied by a character portion composed of one or more groups of characters input by handwriting, for example. In such a case, the feature amount is the ratio of the area occupied by the character portion composed of one or more groups. The character portion is composed of one or more groups.

  The feature amount acquisition unit 14323 calculates, for example, the distance between each group of characters input by handwriting as the feature amount. In such a case, the feature amount is a distance between the groups. The feature amount acquisition unit 14323 calculates, for example, the shortest distance between each group of input characters and each other group. Then, the feature quantity acquisition unit 14323 calculates, for example, the sum of the obtained distances. In such a case, the feature amount is the sum of the distances between one or more groups.

  The feature amount acquisition unit 14323 acquires the length of a specific line segment such as a honey, for example. In such a case, the feature amount is the length of a specific line segment.

  The feature amount acquisition unit 14323 can be usually realized by an MPU, a memory, or the like. The processing procedure of the feature amount acquisition unit 14323 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The balance determination unit 14324 determines whether or not the character is an incorrect answer using one or more feature amounts acquired by the feature amount acquisition unit 14323. The balance determination unit 14324 applies one or more feature amounts acquired by the feature amount acquisition unit 14323 to the determination information, and determines whether or not a condition indicated by the determination information is satisfied. When the condition indicated by the determination information is not satisfied, the balance determination unit 14324 determines that the input character is incorrect. Note that the determination information may be only a feature amount, or may be a rule or expression for determination having a feature amount. The determination information is usually included in the correct character information in the correct character information storage unit 11. The balance determination unit 14324 acquires, for example, determination information including a ratio of the area occupied by a character portion composed of one or more correct character groups, and the feature amount and determination information calculated by the feature amount acquisition unit 14323. From this, it is determined whether or not the character is incorrect. For example, the balance determination unit 14324 acquires determination information including the distance between groups of correct characters, and determines whether the character is an incorrect answer from the feature amount calculated by the feature amount acquisition unit 14323 and the determination information. .

  The balance determination unit 14324 can be usually realized by an MPU, a memory, or the like. The processing procedure of the balance determination unit 14324 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The determination result output unit 15 outputs a determination result. Further, the determination result output unit 15 may output the determination result in the determination unit 14, the correct character string, and the input character image. The determination result may be a correct answer when comparing two or more characters and when all the characters are correct. The determination result may include a character image in addition to information indicating a correct answer or an incorrect answer. It is preferable that the determination result output unit 15 outputs the determination result and the determination reason. Further, it is preferable that the determination result output unit 15 outputs the determination reason only when it is determined that the answer is incorrect. The reason for determination is, for example, a comment to be described later. The comment is information included in the determination information. Here, output refers to display on a display, projection using a projector, printing on a printer, sound output, transmission to an external device, storage in a recording medium, output to another processing device or other program, etc. It is a concept that includes delivery of processing results. The output may be two or more different processes such as display on a display and storage on a recording medium.

  The determination result output unit 15 may or may not include an output device such as a display or a speaker. The determination result output unit 15 can be implemented by output device driver software, or output device driver software and an output device.

  The receiving unit 16 receives the user determination result after the determination result output unit 15 outputs the determination result. The user determination result is a user determination result indicating whether the character string input by handwriting is correct or incorrect. In addition, the reception unit 16 may receive a user determination result and a determination reason. The determination reason may be accepted only when a certain condition is satisfied. The certain condition is that the determination result in the determination unit 14 is different from the user determination result received by the reception unit 16, or the determination result in the determination unit 14 is ○ (OK) and the user determination result is × (NG). And so on. The receiving unit 16 may receive a feature amount or determination information including the feature amount. Here, reception means reception of information input from an input device such as a keyboard, mouse, touch panel, reception of information transmitted via a wired or wireless communication line, recording on an optical disk, magnetic disk, semiconductor memory, or the like. It is a concept including reception of information read from a medium. The user judgment result input means may be anything such as a numeric keypad, keyboard, mouse or menu screen. The accepting unit 16 can be realized by a device driver for input means such as a numeric keypad and a keyboard, control software for a menu screen, and the like.

  The determination result accumulation unit 17 accumulates the user determination result received by the reception unit 16. The determination result storage unit 17 preferably stores the user determination result received by the receiving unit 16 and the determination reason. The determination result accumulation unit 17 can be usually realized by an MPU, a memory, or the like. The processing procedure of the determination result storage unit 17 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The determination information update unit 18 updates the determination information when the determination result in the determination unit 14 and the user determination result received by the reception unit 16 satisfy a predetermined condition. The determination information update unit 18 may update the determination information when the determination result is different from the user determination result. In addition, the determination information update unit 18 may update the determination information when the determination result is “incorrect (NG)” and the user determination result is “correct (OK)”. Further, the determination information update unit 18 may update the determination information when the determination result is “correct (OK)” and the user determination result is “incorrect (NG)”. Furthermore, the determination information update unit 18 may update the determination information using one or more feature amounts received by the reception unit 16. The determination information is information serving as a reference for determining whether or not a character included in the input character image is an incorrect answer. The determination information is information indicating a condition having a feature amount, for example.

  The judgment information update unit 18 can be usually realized by an MPU, a memory, or the like. The processing procedure of the determination information updating unit 18 is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

  The comparison output unit 19 compares the determination result output from the determination result output unit 15 with the user determination result received by the reception unit 16, and outputs a determination reason when both results are different. When the determination result and the user determination result are different, the comparison output unit 19 preferably outputs both the determination result and the user determination result in addition to the determination reason. Note that when both results are different, the determination result is “correct” and the user determination result is “incorrect”, or the determination result is “incorrect” and the user determination result is “correct”. Is the case.

  In addition, the comparison output unit 19 has both the determination result output by the determination result output unit 15 and the user determination result received by the reception unit 16 being “incorrect”, and the determination reason output by the determination result output unit 15 ( When the comment) and the determination reason input by the user are different, it is preferable to output the determination reason. In such a case, it is preferable that the comparison output unit 19 outputs both the determination result and the user determination result in addition to the determination reason. The comparison output unit 19 performs a process of comparing the determination reason (comment) output from the determination result output unit 15 with the determination reason input by the user. For example, this process includes a determination reason (comment) that is a character string. ) And the determination reason entered by the user as a character string, if one or more independent words are acquired and one or more independent words acquired from both character strings match, it is determined that the determination reasons are the same If one or more independent words do not match, it may be determined that the reason for determination is different. Also, a code is assigned to the judgment reason (comment) and the judgment reason entered by the user. If both codes match, the judgment reason is judged to be the same, and if both codes do not match, the judgment reason May be determined to be different.

  Here, output refers to display on a display, projection using a projector, printing on a printer, sound output, transmission to an external device, storage in a recording medium, output to another processing device or other program, etc. It is a concept that includes delivery of processing results. The output may be two or more different processes such as display on a display and storage on a recording medium. The comparison output unit 19 may or may not include an output device such as a display or a speaker. The comparison output unit 19 can be realized by MPU and output device driver software, or MPU and output device driver software and output device.

  Next, operation | movement of the scoring apparatus 1 is demonstrated using the flowchart of FIG.

  (Step S601) The scoring device 1 performs an automatic scoring process. The automatic scoring process will be described using the flowchart of FIG.

  (Step S602) The scoring device 1 performs a manual scoring process. The operation of the manual scoring process will be described with reference to the flowchart of FIG. The outline of the manual scoring process is as follows. First, the user looks at the determination result output by the determination result output unit 15. Here, the determination result usually includes the answer (character image) of the subject (the person who took the test). And it is suitable for the judgment result output part 15 to output the correct character further. A correct character is a character string composed of correct character information. Next, the user determines whether or not the test subject's answer (character image) is correct. And a user inputs a user judgment result. Next, the reception unit 16 receives a user determination result. In this case, the reception unit 16 may receive a feature amount or the like for changing the determination information. Next, the determination result storage unit 17 stores the user determination result in association with the determination result. Further, the user may input a reason for determination. In such a case, the reception unit 16 also receives a determination reason. Then, the determination result storage unit 17 stores the user determination result in association with the determination result and the determination reason.

  (Step S603) The scoring device 1 performs post-processing. Post-processing will be described with reference to the flowchart of FIG. The process ends. The post-process is a process for reflecting the user's judgment in the automatic scoring process.

  In the flowchart of FIG. 6, automatic scoring and manual scoring may be performed sequentially for each question.

  Next, the operation of the automatic scoring process in step S601 will be described using the flowchart of FIG.

  (Step S701) The character image reading unit 10 substitutes 1 for the counter i.

  (Step S702) The character image reading unit 10 determines whether or not the i-th answer exists. If the i-th answer exists, the process proceeds to step S703, and if the i-th answer does not exist, the process returns to the upper process.

  (Step S703) The character image reading unit 10 determines whether one or more characters input by handwriting have been read. If the character is read, the process goes to step S705, and if the character is not read, the process returns to step S703. The character image read here is stored in the character image storage unit 12.

  (Step S704) The character dividing unit 141 acquires a character image from the character read in step S703.

  (Step S705) The character dividing unit 141 divides the character image acquired in step S704 into an image of each character.

  (Step S706) The point information acquisition means 142 substitutes 1 for the counter i.

  (Step S707) The point information acquisition unit 142 determines whether or not the i-th character image exists in the character image divided in step S705. If the i-th character image exists, the process goes to step S708. If the i-th character image does not exist, the process goes to step S712.

  (Step S708) The point information acquisition unit 142 performs point information acquisition processing. A point information acquisition process is demonstrated using the flowchart of FIG.

  (Step S709) The determination unit 143 performs determination processing. The determination process will be described with reference to the flowchart of FIG.

  (Step S710) The point information acquisition unit 142 determines whether the result of the determination process in step S709 is “OK” or “NG”. If “OK”, go to step S711, and if “NG”, go to step S712.

  (Step S711) The point information acquisition unit 142 increments the counter i by 1. The process returns to step S707.

  (Step S712) The determination result output unit 15 outputs a determination result. Note that the determination result output unit 15 preferably outputs the input character image and the determination result. In addition to the determination result, it is preferable that the determination result output unit 15 outputs a correct character composed of correct character information.

  (Step S713) The character image reading unit 10 increments the counter i by one. The process returns to step S702.

  Next, the point information acquisition processing in step S706 will be described using the flowchart in FIG.

  (Step S801) The point information acquisition unit 142 removes the thickness from the handwritten character image and converts it to a line drawing image. Such processing is a known technique.

  (Step S802) The point information acquisition unit 142 converts the line drawing image obtained in step S801 into line drawing data (vector) to obtain vector data. Such processing is a known technique.

  (Step S803) The structure point information acquisition unit 1421 extracts a structure point from the vector data obtained in step S802. Further, the feature point information acquisition unit 1422 extracts feature points from the vector data obtained in step S802. The method for extracting structure points and feature points is as described above.

  (Step S804) The structure point information acquisition unit 1421 or the feature point information acquisition unit 1422 acquires the position information (here, coordinate information (x, y)) of the structure point or the feature point and accumulates it in the storage medium.

  (Step S805) The structure point information acquisition unit 1421 or the feature point information acquisition unit 1422 acquires the connection information of the structure point and the feature point and accumulates them in the storage medium. Return to upper process.

  In the flowchart of FIG. 8, the extraction of structure points and feature points, acquisition of position information, and acquisition of connection information may not be performed sequentially but may be performed in parallel.

  Next, the determination process in step S707 will be described with reference to the flowchart of FIG.

  (Step S <b> 901) The determination unit 143 reads the correct character information of the correct character corresponding to the input character to be determined from the correct character information storage unit 11.

  (Step S902) The structure determining unit 1431 performs a structure comparison process. The structure comparison process will be described with reference to the flowchart of FIG.

  (Step S903) The feature determination unit 1432 determines whether or not the determination result in step S902 is “OK”. If it is “OK”, the process goes to step S904, and if it is not “OK”, the process returns to the upper process.

  (Step S904) The feature determination unit 1432 performs a shape comparison process. The shape comparison process will be described with reference to the flowchart of FIG.

  (Step S905) The feature determination unit 1432 determines whether or not the determination result in step S904 is “OK”. If it is “OK”, the process goes to step S906, and if it is not “OK”, the process returns to the upper process.

  (Step S906) The feature determination unit 1432 performs a feature amount comparison process. The feature amount comparison process will be described with reference to the flowchart of FIG.

  In the flowchart of FIG. 9, it is preferable to perform the structure comparison process first, but the order of the structure comparison process, the shape comparison process, and the feature amount comparison process does not matter.

  Next, the structure comparison process in step S902 will be described with reference to the flowchart of FIG.

  (Step S1001) The structure determination means 1431 divides the structure points and feature points included in one character into one or more groups having connected structure points or feature points. The process of dividing into groups may be performed by the group dividing unit 14321.

  (Step S1002) The structure determining unit 1431 performs an initialization process. The initialization process is a process of substituting 1 for the counter i and substituting “OK” for the variable determination result.

  (Step S1003) The structure determining unit 1431 determines whether or not the i-th group exists. If the i-th group exists, the process goes to step S1003, and if the i-th group does not exist, the process returns to the upper process.

  (Step S1004) The structure determination means 1431 acquires the structure point information of the i-th group of input characters.

  (Step S1005) The structure determination means 1431 acquires the structure point information of the i-th group of correct characters.

  (Step S1006) Structure determination means 1431 substitutes 1 for counter j.

  (Step S1007) The structure determining unit 1431 determines whether or not the j-th structure point exists in the i-th group of input characters. If the jth structural point exists, the process proceeds to step S1008, and if the jth structural point does not exist, the process proceeds to step S1010.

  (Step S1008) The structure determination means 1431 determines whether or not the structure point corresponding to the j-th structure point exists in the i-th group of correct character information. If it exists, go to step S1009, otherwise go to step S1011. Whether or not it is “corresponding” can be determined by, for example, leaving only the node of the structure point from the connection information of the i-th group and inspecting the remaining node. The inspection is to determine whether or not there is a structure point corresponding to the correct character in the order of connection of the structure points of the input character. It is preferable to determine whether or not it is compatible only with the connection information of the structure point, but it may be determined only with the position information of the structure point, or both the connection information and the position information of the structure point may be determined. Judgment may be made by using.

  (Step S1009) Structure determination means 1431 increments counter j by 1. The process returns to step S1007.

  (Step S1010) The structure determining unit 1431 determines whether or not there is a structure point (remaining structure point) that is not the object of determination in determining whether or not to correspond in step S1008. If there is a surplus structure point, the process goes to step S1011. If there is no surplus structure point, the process goes to step S1012.

  (Step S1011) Structure determination means 1431 substitutes “NG” for the determination result. Return to upper process.

  (Step S1012) Structure determination means 1431 increments counter i by one. The process returns to step S1003.

  Next, the shape comparison process of step S904 will be described using the flowchart of FIG.

  (Step S1101) The group dividing means 14321 divides the structure points and feature points included in one character into one or more groups having connected structure points or feature points.

  (Step S1102) The group feature determination unit 14322 performs an initialization process. The initialization process is a process of substituting 1 for the counter i and substituting “OK” for the variable determination result.

  (Step S1103) The group feature determination unit 14322 determines whether or not the i-th group exists. If the i-th group exists, the process goes to step S1104, and if the i-th group does not exist, the process returns to the upper process.

  (Step S1104) The group feature determination unit 14322 acquires the structure point information and the feature point information of the i-th group of the input character.

  (Step S1105) The group feature determination means 14322 acquires the structure point information and feature point information of the i-th group of correct characters.

  (Step S1106) The group feature determination means 14322 compares the number of nodes of the i-th group of input characters and the i-th group of correct characters, and determines the group of characters having the larger number of nodes. The following check is performed based on the group of characters with the larger number of nodes.

  (Step S1107) The group feature determination unit 14322 performs an initialization process. The initialization process is a process of substituting 1 for the counter j and substituting “0” for the evaluation value of the variable.

  (Step S1108) The group feature judging means 14322 judges whether or not the jth node of the reference group exists. If there is a node, the process goes to step S1109, and if there is no node, the process goes to step S1113.

  (Step S1109) The group feature determining unit 14322 determines a node corresponding to the j-th node (reference character node) among the nodes of the other character group that is not the reference, and sets a pair of two nodes. Information is stored at least temporarily. Note that the node corresponding to the j-th node is, for example, the node closest to the j-th node. A pair of two nodes is two nodes corresponding to two characters.

  (Step S1110) The group feature determination unit 14322 calculates the distance between the two nodes that were paired in step S1109. The group feature determination unit 14322 calculates a distance from two pieces of position information included in information of two nodes.

  (Step S1111) The group feature determination unit 14322 adds a distance to the evaluation value. The evaluation value is finally the sum of the distances between the two nodes.

  (Step S1112) The group feature judging means 14322 increments the counter j by 1. The process returns to step S1108.

  (Step S1113) The group feature determination unit 14322 determines whether or not the evaluation value is greater than a threshold value (including whether or not the evaluation value is greater than or equal to). If so, go to step S1114, otherwise go to step S1115.

  (Step S1114) The group feature determination means 14322 substitutes “NG” for the determination result. Return to upper process.

  (Step S1115) The group feature judging means 14322 increments the counter i by 1. The process returns to step S1103.

  Next, the feature amount comparison processing in step S906 will be described using the flowchart in FIG. The feature amount comparison process includes a balance comparison process.

  (Step S1201) The feature quantity acquisition unit 14323 acquires the rectangular area of the input character. The area of the rectangle of the input character is, for example, the area of the rectangle including all the nodes of the input character.

  (Step S1202) The feature amount acquisition unit 14323 performs initialization processing. The initialization process is a process of substituting 1 for the counter i and substituting “OK” for the determination result.

  (Step S1203) The feature quantity acquisition unit 14323 determines whether or not the i-th group exists in the input character. If the i-th group exists, the process goes to step S1204. If the i-th group does not exist, the process goes to step S1211.

  (Step S1204) The feature amount acquisition unit 14323 calculates the rectangular area of the i-th group.

  (Step S1205) The feature quantity acquisition unit 14323 calculates the ratio of the area of the i-th group to the area of the entire character.

  (Step S1206) The balance determination unit 14324 determines whether or not the ratio acquired in step S1205 is within an allowable range. The allowable range is determination information, and is stored in the correct character information storage unit 11, for example. If it is within the allowable range, the process goes to step S1207, and if it is not within the allowable range, the process goes to step S1210. Note that the information regarding the allowable range is, for example, determination information included in the correct character information.

  (Step S1207) The feature quantity acquisition unit 14323 calculates the distance between the i-th group and another group. The distance may be the distance between the center of gravity of the rectangle of the i-th group and the center of gravity of the rectangle of the other group, or the closest two nodes among the distances between the nodes of the i-th group and the nodes of the other group It may be the distance between. Here, when there are three or more groups, the feature amount acquisition unit 14323 may calculate two or more distances. The other group may be one adjacent group, one or more adjacent groups, or all other groups.

  (Step S1208) The balance determination unit 14324 determines whether or not the value calculated using the one or more distances acquired in Step S1207 is within an allowable range. If it is within the allowable range, go to step S1209, and if it is not within the allowable range, go to step S1210. The value calculated using one or more distances is, for example, the sum of one or more distances. Note that the information regarding the allowable range is, for example, determination information included in the correct character information.

  (Step S1209) The feature quantity acquisition unit 14323 increments the counter i by 1. The process returns to step S1203.

  (Step S1210) The balance determination means 14324 substitutes “NG” for the determination result. Return to upper process.

  (Step S1211) The feature quantity acquisition unit 14323 determines whether there is another feature quantity. If another feature quantity exists, the process proceeds to step S1212, and if another feature quantity does not exist, the process returns to the upper process. Note that whether or not another feature amount exists can be determined by checking the correct character information, for example. In addition, the value corresponding to the other feature amount is, for example, the length of splash. The length of the splash is usually the distance between two specific structural points.

  (Step S1212) The feature amount acquisition unit 14323 acquires a value corresponding to another feature amount from the structure point information of the input character, the feature point information, or the structure point information and the feature point information.

  (Step S1213) The balance determining unit 14324 determines whether or not the value corresponding to the other feature amount acquired in Step S1212 is within an allowable range. If it is within the allowable range, the process goes to step S1211, and if it is not within the allowable range, the process goes to step S1210. Note that the information regarding the allowable range is, for example, determination information included in the correct character information.

  Next, the operation of the manual scoring process in step S602 will be described using the flowchart of FIG.

  (Step S1301) The determination result output unit 15 substitutes 1 for the counter i.

  (Step S1302) The determination result output unit 15 determines whether or not the i-th answer (character image) exists. If the i-th answer exists, the process goes to step S1303, and if it does not exist, the process returns to the upper process.

  (Step S1303) The determination result output unit 15 displays the character image corresponding to the i-th answer and the determination result on the display.

  (Step S1304) The reception unit 16 determines whether a user determination result has been received. If a user determination result is accepted, the process goes to step S1305, and if not, the process returns to step S1304.

  (Step S1305) The determination result storage unit 17 stores the user determination result received in step S1304 in association with the determination result.

  (Step S1306) The receiving unit 16 determines whether or not the user determination result satisfies a predetermined condition. If the predetermined condition is satisfied, the process goes to step S1307. If the predetermined condition is not satisfied, the process goes to step S1310.

  (Step S1307) The determination result output unit 15 performs an output prompting the input of the determination reason. The determination result output unit 15 displays, for example, “Please input the reason for the current determination”. Information for prompting the input of the determination reason is held in advance by the determination result output unit 15.

  (Step S1308) The reception unit 16 determines whether a determination reason has been received. If the determination reason is accepted, the process goes to step S1309, and if not accepted, the process returns to step S1308.

  (Step S1309) The determination result storage unit 17 stores the determination reason received in step S1308 in association with the user determination result.

  (Step S1310) The judgment result output unit 15 increments the counter i by one. The process returns to step S1302.

  Next, the post-processing operation of step S603 will be described using the flowchart of FIG.

  (Step S1401) The comparison output unit 19 substitutes 1 for the counter i.

  (Step S1402) The comparison output unit 19 determines whether or not the i-th answer (character image) exists. If the i-th answer exists, the process goes to step S1403, and if not, the process goes to step S1408.

  (Step S1403) The comparison output unit 19 acquires the i-th determination result corresponding to the i-th answer and the i-th user determination result. The i-th determination result is information acquired by the automatic scoring process. The i-th user determination result is a determination result input by the user.

  (Step S1404) The comparison output unit 19 determines whether or not the i-th determination result matches the i-th user determination result. If they match, go to step S1407, otherwise go to step S1405.

  (Step S1405) The comparison output unit 19 acquires the i-th input character image corresponding to the i-th answer.

  (Step S1406) The comparison output unit 19 acquires a determination reason corresponding to the i-th user determination result.

  (Step S1407) The comparison output unit 19 increments the counter i by one. The process returns to step S1402.

  (Step S1408) The comparison output unit 19 reads correct character information.

  (Step S1409) The comparison output unit 19 displays correct character information (correct character string), input character image, determination reason, and determination information. It is assumed that the determination information is included in the correct character information read in step S1408.

  (Step S1410) The reception unit 16 determines whether or not the update of the feature amount has been received. If the update of the feature amount is accepted, the process goes to step S1411. If the update is not accepted, the process goes to step S1414.

  (Step S1411) The determination information update unit 18 updates the determination information included in the correct character information using the feature amount received by the reception unit 16.

  (Step S1412) The reception unit 16 determines whether an instruction to perform a simulation has been received. If an instruction to perform simulation is accepted, the process goes to step S1413. If not accepted, the process goes to step S1414.

  (Step S1413) An automatic scoring process is performed. The automatic scoring process has been described using the flowchart of FIG. However, the target of the automatic scoring process here is only the input character image displayed in step S1409. Such an input character image is an input character image in which the determination result does not match the user determination result. Further, the automatic scoring process here may be only the process after step S704. Then, by this automatic scoring process, the user can inspect whether or not the judgment information is appropriate by automatically scoring with the judgment condition changed.

  (Step S1414) The reception unit 16 determines whether an instruction to end the process has been received. If an instruction to end the process is accepted, the process returns to the upper process, and if not accepted, the process returns to step S1410.

  Hereinafter, a specific operation of the scoring device in the present embodiment will be described.

  Now, the correct character information storage unit 11 holds the correct character information management table shown in FIGS. 15 and 16. The correct character information management table holds correct character information of one or more correct characters. The correct character information includes structure point information and feature point information. In FIG. 15, the structure point information and the feature point information are node position information and connection information, and are not distinguished. However, in FIG. 15, since end points, branch points, and intersections can be determined from the connection information, structure points and feature points can be distinguished. In FIG. 15, correct character information includes a character code, node position information, and connection information. The pictograph composed of the characters on the right side of FIG. 15 and the numbers in the circles is a pictograph showing the image of the character composed of the node position information and the connection information. The circle on the right side of FIG. 15 indicates a structure point or a feature point. The structure point is a white circle. The feature points are shaded circles. In the character “ka”, P1 of the node position information indicates the structure point of the circle 1 of the right pictogram. In the character “ka”, the node position information P3 indicates the feature point of the circle 3 of the pictogram on the right side. Since the character has a normalized size, the node position information falls within the range of (0, 1). The node position information is node position information. The connection information is information indicating connection of points, and the data in FIG. 15 is an example. Further, in FIG. 15, a flag indicating the type of node (structure point or feature point) may be provided corresponding to the node position information.

  In FIG. 16, the correct character information management table holds correct character information of one or more correct characters. The correct character information includes a character code, node position information, connection information, and a feature amount. Here, the character is “NA”, and the correct character information has two feature amounts and a comment. In FIG. 16, the correct character information management table has the same structure as that of FIG.

  In FIG. 16, the feature quantity (1) “5 <= P15−P16” indicates that the lower part of the Chinese character “NA” is 5 pixels or more. The comment “insufficiency of honey” corresponding to the feature value (1) in FIG. 16 is a comment that is output when it is determined that the feature value (1) is applied and the character is incorrect.

  In the feature quantity (2) “0.8 <= A / B <= 1.2” in FIG. 2 or less. Note that “A (P1-P4-...)” Of the feature quantity (2) in FIG. 14 indicates the upper part “large” of the Chinese character “na”. “B (P10-P11, P12-...)” Of the feature quantity (2) in FIG. 14 indicates the lower part “indication” of the Chinese character “NA”. The comment “Bad balance” corresponding to the feature value (2) in FIG. 14 is a comment that is output when it is determined that the feature value (2) is applied and the character is incorrect.

  Needless to say, the method of defining feature quantities is not limited to the above.

  In such a situation, it is assumed that a kanji test is performed on a large number of subjects. It is assumed that a number of answer sheets written by a large number of subjects are scored by using the scoring device 1 in the case of a correct answer.

  That is, the scoring device 1 first performs automatic scoring as follows. First, the character image reading unit 10 sequentially reads such answer sheets to obtain a character image. Here, the character image reading unit 10 is, for example, a scanner with an auto feeder.

  Next, the character dividing unit 141 acquires a character image from the read character. Here, since the character string of the input image is one character, the character dividing unit 141 does not perform the dividing process, and acquires the character image of the Chinese character “NA”.

  First, the point information acquisition unit 142 performs the point information acquisition process described above for the Chinese character “NA” of each answer sheet. And the point information acquisition means 142 acquires structure point information and feature point information.

  Next, the determination unit 143 performs the above-described determination process using the structure point information and feature point information acquired by the point information acquisition unit 142. That is, first, the determination unit 143 reads the correct character information (FIG. 16) of the correct character corresponding to the input character to be determined from the correct character information storage unit 11. Next, the structure determination unit 1431 performs the structure comparison process described above.

  Next, if the determination result that is the result of the structure comparison process is “OK”, the feature determination unit 1432 performs a shape comparison process.

  If the determination result, which is the result of the shape comparison process, is also “OK”, the balance comparison process using the following feature amount is performed. That is, the balance determination unit 14324 reads the feature value (1) and the feature value (2) of the Chinese character “NA” from the correct character information storage unit 11.

  Next, the feature amount acquisition unit 14323 acquires the length of the input character (the distance of “P15-P16”) from “P15-P16” of the feature amount (1). Then, the balance determination unit 14324 determines whether the length of the honey acquired by the feature amount acquisition unit 14323 is 5 pixels or more. If it is less than 5 pixels, the balance determination means 14324 will make the input character an incorrect answer.

  Next, when the honey length acquired by the feature amount acquisition unit 14323 is 5 pixels or more, the feature amount acquisition unit 14323 determines from “A = (P1-P4...)” Of the feature amount (2). The area (A) of the “large” rectangle above “N” is acquired. Next, the feature amount acquisition unit 14323 obtains the rectangular area (B) of “shown” below “NA” from “B = (P10−P11, P12−...)” Of the feature amount (2). get. Then, the feature quantity acquisition unit 14323 calculates the ratio (A / B).

  Then, the balance determination unit 14324 determines whether or not the ratio (A / B) satisfies “0.8 <= A / B <= 1.2”. If not satisfied, the balance determination unit 14324 makes the input character an incorrect answer.

  Then, the determination result output unit 15 accumulates the determination result that is the result of the automatic determination in the determination result storage unit 13. The determination result is usually stored in association with an answer sheet identifier (same meaning as the subject identifier) for identifying the answer sheet. Further, it is preferable that the determination result output unit 15 accumulates comments (reasons for determination) in pairs with the determination results. The determination reason is information included in the determination information.

  Next, the user inputs an instruction for causing the display to display a determination result that is a result of automatic determination of the test subject's answer sheet (character image). And the judgment result output part 15 displays a test subject's answer sheet (character image) and a judgment result (refer FIG. 17) in order on a display. In FIG. 17, when the determination result is “× (incorrect answer)”, a comment is output.

  Next, it is assumed that the user (scorer) inputs “◯” or “x” in the field (attribute value) of “user judgment result”. Then, it is assumed that the user (scorer) has input a user determination result as shown in FIG.

  And the reception part 16 receives the user judgment result shown in FIG. Next, the determination result storage unit 17 stores the received user determination result (FIG. 18) in association with the determination result.

  Here, the reception unit 16 determines whether or not the user determination result satisfies a predetermined condition (here, “the determination result is different from the user determination result”). Then, it is determined that a certain condition is satisfied for the answer sheet identifier “002”.

  Next, the determination result output unit 15 performs an output that prompts input of a determination reason. That is, for example, the determination result output unit 15 outputs “Please input the reason for determining“ × ”for the answer sheet identifier“ 002 ”” on the display ”.

  Next, it is assumed that the user inputs “this level of imbalance is an allowable range”. Then, the reception unit 16 receives a determination reason “this level of imbalance is an allowable range”.

  Next, the determination result storage unit 17 stores the received determination reason “this level of imbalance is an allowable range” in association with the answer sheet identifier “002”.

  The specific example of the manual scoring process has been described above.

  Next, a specific example of post-processing will be described.

  First, the determination result output unit 15 acquires a determination result corresponding to each answer sheet and a user determination result. Then, the determination result output unit 15 determines whether the determination result matches the user determination result. If the determination result does not match the user determination result, the correct character information, the input character image that does not match, the determination reason, and the determination information are acquired and output. Examples of such output are shown in FIGS. 19 and 20.

  FIG. 19 shows that the judgment result of the automatic judgment process and the user judgment result in the manual scoring process are different from each other regarding whether or not the reflection is insufficient due to the application of the judgment information including the feature quantity (1). (Scoring target) is displayed.

  Then, it is assumed that the circle of the honey portion in FIG. 19A is drawn by the determination result output unit 15 from “P15-P16”, and the circles including the structural points P15 and P16 are drawn.

  Then, the determination result output unit 15 displays a determination information update screen as shown in FIG. Here, the determination information update screen is a screen for changing the condition (determination information) of the length of the honey. Then, the user changes “5 pixels or more” to “3 pixels or more” on the determination information update screen in FIG. 19B using an input means such as a keyboard, and presses the “update” button. , And. Then, the reception unit 16 receives the correct / incorrect information “3”. Then, the determination information update unit 18 configures the determination information “3 <= P15−P16” from the correct / incorrect information “3” received by the reception unit 16, and the determination information (determination information of the feature (1)) of FIG. ).

  Next, it is assumed that the user presses the “simulation” button in FIG. Then, the feature amount acquisition unit 14323 and the balance determination unit 14324 determine whether each character to be scored satisfies the determination information “3 <= P15−P16” of the feature amount (1) by the above-described algorithm. To do. Then, the determination result output unit 15 outputs a determination result (simulation screen) as shown in FIG. In the determination result of FIG. 20, ◯ indicates a character that is correct by changing the determination information to “3 <= P15−P16”. In the determination result of FIG. 20, × indicates a character that is an incorrect answer even when the determination information is changed to “3 <= P15−P16”. As shown in FIG. 20, the user can obtain appropriate judgment information by performing a simulation.

  FIG. 20 shows that characters (scoring targets) with different judgment results in the automatic judgment processing and user judgment results in the manual scoring processing are displayed regarding the quality of the balance by applying the judgment information including the feature amount (2). ing.

  In order to obtain such a display, the following processing is performed. That is, the feature amount acquisition unit 14323 acquires the area (A) of the “large” rectangle above “N” from “A = (P1-P4...)” Of the feature amount (2). Next, the feature amount acquisition unit 14323 obtains the rectangular area (B) of “shown” below “NA” from “B = (P10−P11, P12−...)” Of the feature amount (2). get. Then, the feature quantity acquisition unit 14323 calculates the ratio (A / B).

  Then, the balance determination unit 14324 determines whether or not the ratio (A / B) satisfies “0.8 <= A / B <= 1.2”. If not satisfied, the balance determination unit 14324 makes the input character an incorrect answer.

  Then, the determination result output unit 15 outputs a determination result as shown in FIG. FIG. 21 (a) outputs characters with different judgment results and user judgment results regarding the judgment information of the feature quantity (2).

  Then, the determination result output unit 15 outputs a determination information update screen as shown in FIG. Here, the determination information update screen is a screen for changing the determination information regarding the ratio (A / B). Then, the user changes “ratio 0.8 to 1.2” to “ratio 0.7 to 1.4” on the determination information update screen in FIG. ”And press the“ Update ”button. Then, the reception unit 16 receives the correct / incorrect information “0.7 or more and 1.4 or less”. Then, the determination information update unit 18 configures the determination information “0.7 <= A / B <= 1.4” from the correct / incorrect information “0.7 to 1.4” received by the receiving unit 16, The judgment information (judgment information of feature quantity (2)) in FIG. 16 is changed.

  Next, it is assumed that the user presses the “simulation” button in FIG. Then, the feature amount acquisition unit 14323 and the balance determination unit 14324 determine that each character to be scored has the determination information “0.7 <= A / B <= 1.4” of the feature amount (2) according to the algorithm described above. It is determined whether or not the above is satisfied. Then, the determination result output unit 15 outputs a determination result (simulation screen) as shown in FIG. In the determination result of FIG. 22, ◯ indicates a character that is correct by changing the determination information to “0.7 <= A / B <= 1.4”. In the determination result of FIG. 22, × indicates a character that is an incorrect answer even when the determination information is changed to “0.7 <= A / B <= 1.4”. As shown in FIG. 22, the user can obtain appropriate determination information by performing a simulation.

  As described above, the simulation is performed by converting the feature amount for characters (answers) having different judgment results and user judgment results. However, for example, the scoring device 1 may perform a simulation on characters that are determined to be incorrect by the automatic scoring process. Moreover, the scoring device 1 may perform a simulation for all characters (answers).

  Next, another specific example of the automatic scoring process will be described. First, it is assumed that the character image reading unit 10 has read a character “POSTO” inputted by handwriting by the user.

  Next, the character dividing unit 141 acquires a character image from the read character “POSTO”. The character image is a bitmap having a thickness as shown in FIG.

  The character dividing unit 141 divides the acquired character image “POSTO” into three character images “PO”, “SU”, and “TO”.

  Next, the point information acquisition unit 142 performs a point information acquisition process on the first character image “PO” as follows. That is, the point information acquisition unit 142 removes the thickness from the handwritten character image and converts it to a line drawing image (see FIG. 2B). Then, the point information acquisition unit 142 converts the line drawing image into line drawing data (vector) to obtain vector data.

  Next, the structure point information acquisition unit 1421 determines the end points, branch points, and intersections of the vector data as structure points, and accumulates the position information of the structure points.

  Next, the feature point information acquisition unit 1422 acquires feature point information of vector data. That is, the feature point information acquisition unit 1422 acquires an angle between edges included in the vector data, and detects an intersection of edges as a feature point when the degree of bending indicated by the angle is larger than a threshold, and the feature point Is stored.

  Next, the point information acquisition unit 142 acquires and accumulates connection information of structure points and feature points. Through the above processing, unnormalized graph data can be acquired (see FIG. 2C).

  Next, the point information acquisition unit 142 normalizes the position of the structure point and the feature point so that the graph data fits in the rectangle of (0,0) (0,1) (1,1) (1,0). Transform information. Since the normalization process is a known technique, a detailed description thereof is omitted. Through the above processing, normalized graph data can be acquired (see FIG. 2D).

  Next, the determination unit 143 performs the following determination process. That is, the determination unit 143 reads from the correct character information storage unit 11 the correct character information of the correct character (information of “Po” in FIG. 15) corresponding to the input character “Po” to be determined.

  Next, the structure determination unit 1431 performs a structure comparison process as follows. That is, the structure determination unit 1431 divides the structure points and feature points included in the character “Po” into three groups having connected structure points or feature points. Then, according to the flowchart of FIG. 10, the structure determining unit 1431 determines whether or not a node corresponding to each node of the input character “Po” exists in the node of the correct character “Po” for each group. to decide.

  Then, according to the flowchart of FIG. 10, the structure determination unit 1431 has, for all nodes of the input character “Po”, corresponding nodes exist in the node of the correct character “Po”, and the correct character “Po”. If there is no extra node among the “po” nodes, it is determined that the input character “po” is not incorrect (see FIG. 3). A surplus node is a node for which no corresponding node exists.

  Next, the feature determination unit 1432 performs the following shape comparison process (see the flowchart of FIG. 11). That is, the group dividing unit 14321 determines whether or not the shape of the input character is the same as the correct character for each group of the characters “PO”. Specifically, as shown in FIG. 5, the feature determination unit 1432 associates a node constituting the input character and a node constituting the correct character for each group. The feature determining unit 1432 calculates the distance between the input character node and the correct character node (corresponding node). Then, the feature determination unit 1432 calculates an evaluation value using a value of one or more distances. An example of the evaluation value calculation formula is the above-described formula 1.

  Then, the feature determination unit 1432 reads a threshold value stored in advance. Then, the feature determination unit 1432 compares the evaluation value with the threshold value, and determines whether or not the shape of the input character has a degree of approximation higher than a predetermined value compared with the shape of the correct character. If the degree of approximation exceeds a predetermined value, the input character is determined to be incorrect.

  The above processing is performed for all groups, and if even one group does not have a degree of approximation higher than a predetermined value, it is determined that the input character is incorrect.

  Next, the feature amount acquisition unit 14323 and the balance determination unit 14324 perform feature amount comparison processing.

  The above processing is also performed on the second character “SU” and the third character “TO”, and when all of the three characters are not incorrect, it is determined that the input character is correct.

  Then, the determination result output unit 15 outputs the determination result. The output mode of the determination result is not limited.

  As described above, according to the present embodiment, the automatic determination is performed in the determination of the correct / incorrect answer of the handwritten character string, and after the determination result of the automatic determination is output, the grader (user) manually determines. In other words, the grader (user) can score while looking at the automatic determination result, so that it is possible to determine the correctness of the character string very efficiently and accurately. For example, in the case of scoring hundreds of thousands of Kanji tests, if the scoring time can be shortened even by 1 second per mail, the effect will be very great.

  Further, according to the present embodiment, it is possible to appropriately determine whether a character is correct or incorrect. In particular, according to the present embodiment, a structure point and a feature point are extracted from an input character, and the structure point information and the feature point information are used differently (a different algorithm is applied). The right / wrong judgment can be made appropriately.

  Further, according to the present embodiment, first, the scoring device 1 performs the structure comparison using the structure points, and then performs the feature comparison using the structure points and the feature points. Therefore, it was possible to judge different characters at high speed and with high accuracy.

  Moreover, according to this Embodiment, the scoring apparatus 1 checked the shape and characteristic of each group using the structure point and the feature point. For this reason, it is possible to judge the correctness of characters with high accuracy.

  Further, according to the present embodiment, the scoring device 1 further checks the characteristic points of characters (such as honey and balance) using the feature amount. For this reason, it is possible to judge the correctness of characters with extremely high accuracy.

  Furthermore, according to the present embodiment, the scoring device 1 can change the feature amount (or judgment information) while observing one or more scored input characters, and can simulate. Therefore, the user can set an appropriate feature amount (or determination information) while viewing the actually input characters.

  In the present embodiment, the determination information update unit 18 updates the determination information using the feature amount input by the user. However, the determination information update unit 18 may update the determination information using the user determination result input by the user. That is, when the determination result is different from the user determination result, the determination information update unit 18 respects the user determination result, and the corresponding feature amount of the character that is the target of the user determination result (such as the length of the honey or the balance). And the determination information may be updated using the feature amount. More specifically, when the determination information is “5 <= P15−P16”, and the user determination result is “OK (correct answer)” for the character having the honey length “4”, the determination information The update unit 18 may correct the determination information to “4 <= P15−P16”. On the other hand, when the determination information is “5 <= P15−P16” and the user determination result is “NG (incorrect answer)” for the character having the length “6”, the determination information update unit 18 May correct the determination information to “6 <P15−P16”.

  In addition, the determination information is “0.8 <= A / B <= 1.2”, and the user determination result is “OK (correct answer)” for the character with “A / B = 0.7”. In some cases, the determination information update unit 18 may correct the determination information to “0.7 <= A / B <= 1.2”. On the other hand, the determination information is “0.8 <= A / B <= 1.2”, and the user determination result is “NG (incorrect answer)” for the character with “A / B = 0.85”. In this case, the determination information update unit 18 may correct the determination information to “0.85 <A / B <= 1.2”.

  In addition, the determination information update unit 18 automatically determines the determination information as described above so as to respect the user determination result only when the determination result is “NG” and the user determination result is “OK”. It may be updated.

  In addition, the determination information update unit 18 automatically determines the determination information as described above so that the user determination result is respected only when the determination result is “OK” and the user determination result is “NG”. It may be updated.

  By the above processing, when the automatic scoring judgment result and the manual scoring judgment result are different, the automatic scoring judgment logic of the scoring device 1 can be adjusted.

  Moreover, in this Embodiment, the scoring apparatus 1 may extract only a structure point, and may perform a right / wrong determination using only a structure point at the time of character correct / incorrect determination. In such a case, it is possible to make a correct / incorrect determination at a very high speed.

  Further, according to the present embodiment, the feature amount (or determination information) is mainly a ratio between the portions of the honey or the character. However, it goes without saying that other feature amounts (or determination information) may be applied. The other feature amount (or determination information) is, for example, a distance between character portions. The character part is one or more groups. The ratio may not be an area ratio. The ratio may be the ratio of the diagonal lengths of the rectangles in the character portion.

  Furthermore, the processing in the present embodiment may be realized by software. Then, this software may be distributed by software download or the like. Further, this software may be recorded and distributed on a recording medium such as a CD-ROM. This also applies to other embodiments in this specification. Note that the software that implements the information processing apparatus according to the present embodiment is the following program. In other words, this program includes one or more correct answers including correct character information indicating correct characters on a recording medium, and structure point information that is information of structure points that are end points, branch points, or intersections constituting the skeleton of the character. Character information and a character image that is an image including one or more characters input by handwriting are stored, the computer compares the character image with the correct character information, and the character string input by handwriting is A determination unit that acquires a determination result indicating whether the answer is correct or incorrect, a determination result output unit that outputs the determination result, and the determination result output unit that outputs the determination result, and then the handwritten input character string is correct A reception unit that receives a user determination result, which is a user determination result indicating whether the answer is incorrect, or a program for functioning as a determination result storage unit that stores the user determination result received by the reception unit , It is.

  In the above program, the reception unit receives a user determination result and a determination reason, and the determination result storage unit causes the computer to function as a storage of the user determination result and the determination reason received by the reception unit. Is preferred.

  In the above program, the correct character information includes a correct character string and determination information that is information for correct / incorrect determination, and the determination result in the determination unit and the user determination result received by the reception unit are When they are different from each other, it is preferable to make the computer function as further including a determination information update unit that updates the determination information.

  In the above program, the determination information includes one or more feature amounts characterizing a correct character, and the determination unit uses the determination information including one or more feature amounts to determine whether a character included in the character image is a character string. It is preferable to make a computer function as a means of determining whether or not the answer is incorrect.

  Further, in the above program, the determination unit detects one or more structure points for each character included in the character image stored in the character image storage unit, and a structure point that is information on the structure point Point information acquisition means for acquiring one or more information, structure point information included in correct character information corresponding to each character included in the character image, and structure point information acquired by the point information acquisition means are compared. It is preferable to cause the computer to function as a judgment means for judging whether or not each character included in the character image is an incorrect answer.

  In the above program, the correct character information is a point on an edge that is a line connecting two structure points and structure point information that is information of a structure point that is an end point, a branch point, or an intersection that constitutes a character skeleton. And feature point information that is feature point information that is a point having a degree of bending larger than a threshold value, and the point information acquisition means includes each of the character images stored in the character image storage unit. Structure point information acquisition means for acquiring one or more structure point information for a character, and one or more feature point information for each character included in the character image stored in the character image storage unit Characteristic point information acquisition means, and the determination means obtains one or more pieces of structure point information acquired by the structure point information acquisition means and the feature point information acquisition by using the structure point information and the feature point information differently. Means acquired The above feature point information is compared with one or more structure point information and one or more feature point information included in the correct character information to determine whether or not the character included in the character image is an incorrect answer. It is preferable to make a computer function.

  In the above program, the determination unit compares the one or more structure point information acquired by the structure point information acquisition unit with one or more structure point information included in the correct character information, and is included in the character image. Structure determination means for determining whether or not a character is an incorrect answer, and one or more acquired by the structure point information acquisition means for a character that the structure determination means determines that a character included in a character image is not an incorrect answer The structure point information and the one or more feature point information acquired by the feature point information acquisition means are compared with one or more structure point information and one or more feature point information included in the correct character information. It is preferable to make a computer function as what comprises the characteristic judgment means which judges whether it is a correct answer.

  Further, in the above program, the feature determining means includes a group dividing means for dividing the structure points and feature points included in one character into one or more groups having connected structure points or feature points, and for each group. One or more structure point information acquired by the structure point information acquisition means, one or more feature point information acquired by the feature point information acquisition means, one or more structure point information included in the correct character information, and one or more Group feature determining means for comparing the feature point information of the character and determining whether the character is incorrect or not, and feature amount acquiring means for acquiring one or more feature amounts indicating a balance between groups of handwritten characters. And a balance judging means for judging whether or not the character is an incorrect answer using one or more feature quantities acquired by the feature quantity acquiring means. It is preferable to ability.

  Further, in the above program, the group feature determination means includes, for each group, one or more structure point information acquired by the structure point information acquisition means and one or more feature point information acquired by the feature point information acquisition means. Corresponding point information corresponding to each point to be indicated, and corresponding point information that is information of a point indicated by either one or more structural point information or one or more feature point information included in the correct character information, The distance between each point indicated by each of the one or more structural point information and the one or more feature point information and each point indicated by the corresponding point information is calculated, and the sum of the absolute values of the distances is calculated. When the sum of the absolute values of is greater than or equal to a threshold value or greater than the threshold value, it is preferable to cause the computer to function as determining that the character is incorrect.

  Further, in the above program, the feature amount acquisition unit calculates a ratio of an area occupied by a character part composed of one or more groups of characters input by handwriting as a feature amount, and the balance determination unit includes a correct answer Judgment information including a ratio of the area occupied by a character part composed of one or more groups of characters is acquired, and whether or not the character is an incorrect answer from the feature amount calculated by the feature amount acquisition unit and the determination information It is preferable to make a computer function as a means for determining whether or not.

  In the above program, the feature amount acquisition unit calculates a distance between each group of handwritten characters as a feature amount, and the balance determination unit includes determination information including a distance between correct character groups. It is preferable to cause the computer to function as what is obtained and judged from the feature quantity calculated by the feature quantity obtaining means and the judgment information whether or not the character is incorrect.

  FIG. 23 shows the external appearance of a computer that executes the program described in this specification and realizes the scoring device of the above-described embodiment. The above-described embodiments can be realized by computer hardware and a computer program executed thereon. FIG. 23 is an overview diagram of the computer system 340, and FIG. 24 is a block diagram of the computer system 340.

  23, a computer system 340 includes a computer 341 including an FD drive and a CD-ROM drive, a keyboard 342, a mouse 343, and a monitor 344.

  24, in addition to the FD drive 3411 and the CD-ROM drive 3412, the computer 341 stores an MPU 3413, a bus 3414 connected to the CD-ROM drive 3412 and the FD drive 3411, and a program such as a bootup program. A RAM 3416 for temporarily storing application program instructions and providing a temporary storage space; and a hard disk 3417 for storing application programs, system programs, and data. Although not shown here, the computer 341 may further include a network card that provides connection to the LAN.

  A program that causes the computer system 340 to execute the functions of the scoring device and the like of the above-described embodiment is stored in the CD-ROM 3501 or FD 3502, inserted into the CD-ROM drive 3412 or FD drive 3411, and further stored in the hard disk 3417. May be forwarded. Alternatively, the program may be transmitted to the computer 341 via a network (not shown) and stored in the hard disk 3417. The program is loaded into the RAM 3416 at the time of execution. The program may be loaded directly from the CD-ROM 3501, the FD 3502, or the network.

  The program does not necessarily include an operating system (OS) or a third-party program that causes the computer 341 to execute the functions of the scoring device according to the above-described embodiment. The program only needs to include an instruction portion that calls an appropriate function (module) in a controlled manner and obtains a desired result. How the computer system 340 operates is well known and will not be described in detail.

  The program does not include processing performed by hardware, for example, processing performed by a modem or an interface card (processing performed only by hardware).

  Further, the computer that executes the program may be singular or plural. That is, centralized processing may be performed, or distributed processing may be performed.

  In each of the above embodiments, each process (each function) may be realized by centralized processing by a single device (system), or by distributed processing by a plurality of devices. May be.

  The present invention is not limited to the above-described embodiments, and various modifications are possible, and it goes without saying that these are also included in the scope of the present invention.

  As described above, the scoring device according to the present invention has an effect that the correctness / incorrectness of characters can be appropriately determined, and is useful as, for example, a scoring device for correctness / incorrectness of Chinese characters.

DESCRIPTION OF SYMBOLS 1 Scoring apparatus 10 Character image reading part 11 Correct character information storage part 12 Character image storage part 13 Judgment result storage part 14 Judgment part 15 Judgment result output part 16 Acceptance part 17 Judgment result storage part 18 Judgment information update part 19 Comparison output part 141 Character division means 142 Point information acquisition means 143 Determination means 1421 Structure point information acquisition means 1422 Feature point information acquisition means 1431 Structure determination means 1432 Feature determination means 14321 Group division means 14322 Group feature determination means 14323 Feature quantity acquisition means 14324 Balance determination means

Claims (12)

A correct character information storage unit capable of storing one or more correct character information having a correct character string and determination information that is information for correct / incorrect determination ;
A character image storage unit capable of storing a character image which is an image including a character string input by handwriting;
A determination unit that compares the character image with the correct character information and obtains a determination result indicating whether the handwritten input character string is correct or incorrect;
A judgment result output unit for outputting the judgment result;
A reception unit that receives a user determination result that is a user determination result indicating whether the character string input by handwriting is correct or incorrect after the determination result output unit outputs the determination result;
A determination result storage unit that stores the user determination result received by the reception unit ;
A scoring apparatus comprising: a determination information update unit configured to update the determination information when a determination result in the determination unit is different from a user determination result received by the reception unit . The reception unit
Accept user judgment results and reasons for judgment,
The determination result accumulation unit
The scoring device according to claim 1, wherein a user determination result and a determination reason received by the reception unit are accumulated. The judgment information is
Having one or more features that characterize the correct character;
The determination unit
Using the determined information comprising one or more feature quantities, grading apparatus of claim 1 or claim 2, wherein characters contained in the character image is determined whether solved fraud. The correct character information is
A part of the one or more feature quantities, having structure point information that is information of a structure point that is an end point, a branch point, or an intersection that constitutes the skeleton of the character;
The determination unit
Point information acquisition means for detecting one or more structure points for each character included in the character image stored in the character image storage unit and acquiring one or more structure point information as information of the structure points When,
The structure point information included in the correct character information corresponding to each character included in the character image is compared with the structure point information acquired by the point information acquisition unit, and whether each character included in the character image is an incorrect answer The scoring device according to any one of claims 1 to 3 , further comprising determination means for determining whether or not. The correct character information is
A point on the edge that is the information of the structure point that is the end point, the branch point, or the intersection of the character skeleton and the line that connects the two structure points. And feature point information which is information of a certain feature point,
The point information acquisition means includes
Structure point information acquisition means for acquiring one or more structure point information for each character included in the character image stored in the character image storage unit;
Feature point information acquisition means for acquiring one or more feature point information for each character included in the character image stored in the character image storage unit;
The determination means includes
Depending on how the structure point information and feature point information are used differently, the one or more structure point information acquired by the structure point information acquisition unit, the one or more feature point information acquired by the feature point information acquisition unit, and the correct character information The scoring device according to claim 4, wherein one or more structure point information and one or more feature point information included in the character image are compared to determine whether or not the character included in the character image is an incorrect answer. The determination means includes
A structure that compares one or more pieces of structure point information acquired by the structure point information acquisition unit with one or more pieces of structure point information included in the correct character information to determine whether or not a character included in the character image is an incorrect answer. Judgment means,
One or more structure point information acquired by the structure point information acquisition unit and one or more acquired by the feature point information acquisition unit for the character that the structure determination unit determines that the character included in the character image is not incorrect claims of comparing the feature point information, one or more of a structural point information and one or more feature point information included in the correct character information, the character includes the feature determining means for determining solved fraud Item 5. The scoring device according to item 5 . The feature judging means includes
Group dividing means for dividing structure points and feature points included in one character into one or more groups having connected structure points or feature points;
For each group, one or more structure point information acquired by the structure point information acquisition unit, one or more feature point information acquired by the feature point information acquisition unit, and one or more structure points included in the correct character information The scoring device according to claim 6 , further comprising group feature judging means for comparing the information with one or more feature point information and judging whether or not the character is an incorrect answer. The feature judging means includes
Feature quantity acquisition means for acquiring one or more feature quantities indicating a balance between groups of handwritten characters;
The scoring device according to claim 7 , further comprising a balance determination unit that determines whether or not the character is an incorrect answer by using one or more feature amounts acquired by the feature amount acquisition unit. The group feature judging means includes
The correct character is a point corresponding to each point indicated by one or more structure point information acquired by the structure point information acquisition unit and one or more feature point information acquired by the feature point information acquisition unit for each group. Obtaining corresponding point information that is information of a point indicated by one of one or more structure point information or one or more feature point information included in the information;
Calculating a distance between each point indicated by each of the one or more structure point information and the one or more feature point information and each point indicated by each corresponding point information;
Calculate the sum of the absolute values of the distances,
The scoring device according to claim 7 or 8 , wherein when the sum of absolute values of the distances is equal to or greater than a threshold value or greater than the threshold value, the character is determined to be incorrect. The scoring device according to claim 1, wherein the character is a Chinese character. On the recording medium,
Storing one or more correct character information having a correct character string and determination information that is information for correct / incorrect determination, and a character image that is an image including a handwritten character string;
A scoring method realized by a determination unit, a determination result output unit, a reception unit, a determination result storage unit, and a determination information update unit,
A determination step in which the determination unit compares the character image with the correct character information and obtains a determination result indicating whether the handwritten input character string is correct or incorrect;
A determination result output step in which the determination result output unit outputs the determination result; and
An accepting step for accepting a user judgment result which is a judgment result of a user indicating whether the handwritten input character string is a correct answer or an incorrect answer after the accepting unit outputs a judgment result in the judgment result output step;
A determination result storage step in which the determination result storage unit stores the user determination result received in the reception step;
A scoring method comprising: a determination information update step in which the determination information update unit updates the determination information when the determination result in the determination step is different from the user determination result received in the reception step. On the recording medium,
Storing one or more correct character information having a correct character string and determination information that is information for correct / incorrect determination, and a character image that is an image including a handwritten character string ;
A computer capable of accessing the recording medium;
A determination unit that compares the character image with the correct character information and obtains a determination result indicating whether the handwritten input character string is correct or incorrect;
A judgment result output unit for outputting the judgment result;
A reception unit that receives a user determination result that is a user determination result indicating whether the character string input by handwriting is correct or incorrect after the determination result output unit outputs the determination result;
A determination result storage unit that stores the user determination result received by the reception unit ;
If the user determination result the receiving unit and the determination result in the determining unit accepts different programs for functioning as a determination information updating unit that updates the determination information.