TWI863841B - Text and graphic element recognition system and method for engineering drawings - Google Patents

Abstract

The present invention relates to a system and method for recognizing text and graphic elements in drawings, particularly suitable for processing rasterized engineering drawings. The system comprises: a frame information recognition unit that recognizes and outputs frame information; an object detection unit that recognizes and outputs object names and corresponding object coordinates; an object parameter detection unit that recognizes and outputs object parameters and object parameter coordinates; an arrow start-end point detection unit that detects and outputs coordinates of arrow start and end points on the drawing; and a parameter attribution detection unit that reads the recognition and detection results of each unit and determines the relationships between various objects and their respective parameters. By employing computer-based automatic image recognition, the invention facilitates the extraction of object shapes and dimensional parameter information, improving accuracy and processing speed. This, in turn, reduces production costs and enhances production efficiency.

Description

System and method for recognizing text and graphic elements in drawings

This invention relates to a recognition system and method, and in particular to a text and graphic element recognition system and method.

Traditional order-based manufacturers often need to input their production parameters into computer numerical control (CNC) processing equipment based on engineering drawings to carry out production. The ideal situation is that the client can provide a general engineering drawing digital file and directly input the file into the CNC processing equipment to obtain the production and processing parameters. However, in most cases, customers only have paper engineering drawings or engineering drawings in portable document format (PDF) or image files. Most manufacturers can only manually fill in the drawing information into the production and processing system for subsequent processing. The processing process is slow and errors caused by human factors are prone to occur during the process, resulting in increased production costs.

Currently, there are technologies related to the application of image processing technology to engineering drawings, such as an engineering drawing image processing device with Taiwan Patent No. M552617, which includes: a storage unit, a processing unit, and a display unit. The storage unit stores color image data of the engineering drawing. The processing unit reads the color image data and performs the following steps: smoothing the color image data with a nonlinear filter; classifying the pixels of the smoothed color image data into crease area pixels and non-crease area pixels with a support vector machine based on the hue and saturation of each pixel in the smoothed color image data; and binarizing the crease area pixels with a regional critical value calculator and binarizing the non-crease area pixels with a global critical value calculator to generate binary image data of the engineering drawing. The display unit displays the binary image of the engineering drawing based on the binary image data. However, the known technology focuses on image processing to improve the clarity of human recognition of image files, so the known technology still needs to rely on human interpretation and input, which is different from the production parameters of the computer numerical control processing equipment that automatically recognizes the drawing in this creation.

Another known technology is a method and system for judging the content of an engineering drawing, such as Taiwan Patent No. I788206, which includes: performing edge detection on the engineering drawing to generate a plurality of initial recognition line segments; performing text recognition on the engineering drawing to generate a plurality of initial recognition texts; providing the initial recognition line segments to a trained structure judgment model for recognition to generate a preliminary recognition result; and performing text recognition on the engineering drawing according to the preliminary recognition result. The method comprises the steps of processing the initial recognition line segments and the initial recognition characters according to the results to generate a plurality of structure line segments and a plurality of structure characters; associating the structure characters with the structure line segments according to a plurality of structure marking properties to generate a plurality of structure information and a plurality of structure labels; generating a plurality of structure recognition results according to the structure information and the structure labels; and outputting the structure recognition results to generate a summary table. However, the known technology is to automatically identify and calculate the amount of beam and column steel bars required for architectural engineering drawings in portable file format files or paper files, and perform engineering structure evaluation calculations based on the drawings. It is unable to automatically identify objects in the engineering drawings and parameters such as the shapes and manufacturing dimensions of the objects, and is unable to input these manufacturing parameters into computer numerical control processing equipment for production.

Therefore, the current known engineering drawing image processing technology has not yet solved the aforementioned problem of being unable to automatically obtain the processing parameters of the paper engineering drawing image to conduct relevant research and judgment on production conditions, or even automatically input the processing parameters into the computer numerical control processing equipment for production.

In view of the above factors, there is a need to develop a system that can directly identify and extract object parameters including specifications, dimensions, tolerances, etc. from images or paper drawings, so as to further directly input the object parameters into computer numerical control processing equipment for automated production.

To solve the above and other problems, the present invention aims to provide a system and method for recognizing text and graphics elements in a drawing to improve the shortcomings of the prior art.

Another purpose of this invention is to provide a system and method for recognizing text and graphic elements in drawings, which can avoid errors caused by human factors in the process of manually filling in drawing information into production equipment, leading to increased production costs.

Another purpose of this invention is to provide a system and method for recognizing text and graphic elements in a drawing, which can use a computer to automatically recognize text, objects and object parameter information in the image, improve the accuracy, reduce production costs and improve production efficiency.

Another purpose of this invention is to provide a drawing text and graphic element recognition system and method, which can identify the size of each part or the type and specification parameters of standard parts from image-based engineering drawings or paper engineering drawings, and automatically input them into computer numerical control processing equipment to automatically produce the parts in the engineering drawings, thereby increasing the proportion of automated production.

In order to achieve the above and other creative purposes, the present creative embodiment provides a drawing text and graphic element recognition system, which is suitable for processing a drawing of a raster engineering drawing, the drawing includes a frame area and a graphic area, the drawing text and graphic element recognition system includes: a frame information recognition unit, in the drawing, according to a preset coordinate, captures a frame image from the frame area , identifying the frame image a frame information; an object detection unit, in the drawing, capturing an object image according to the graphic area, detecting an object width of the object image, segmenting at least one object according to the object width and the frame information, and aggregating an object name and an object coordinate of each object into an object information; an object parameter detection unit, in the drawing In the drawing, a portion corresponding to the object image is deleted to obtain an object parameter map, the object parameter map is identified as an object parameter, and the object parameter and an object parameter coordinate belonging to it are aggregated into an object parameter information; an arrow start and end point detection unit deletes the portion corresponding to the object image and the object parameter map in the drawing to obtain an arrow map, and then detects the arrow. The arrow start and end points in the header image and the corresponding arrow start and end point coordinates are summarized into arrow start and end point information; and a detection parameter attribution unit reads the frame information, the object information, the object parameter information and the arrow start and end point information, determines the relationship between each object and each object parameter by an image algorithm, and summarizes object parameter corresponding information in the form of a text file.

In one embodiment, a production output unit is further included, which imports the object parameter corresponding information into a computer numerical control processing equipment connected to the detection parameter attribution unit information to produce a physical product of the drawing.

In order to achieve the above and other creative purposes, the present creative embodiment provides a method for recognizing text and graphics on a drawing, which is suitable for processing an inputted raster-type engineering drawing, wherein the drawing includes a frame area and a graphic area. The method for recognizing text and graphics on the drawing includes: a frame information recognition step, capturing a frame image according to the frame area, and using an image processing algorithm to recognize the text pattern of the frame image as a frame information; an object detection step, capturing an object image according to the graphic area and identifying a complete object wireframe and an outermost object frame image by an image processing algorithm, reading and scanning the outermost object frame image to segment at least one object according to the frame information, and collecting the object name and an object coordinate as an object information; an object parameter detection step, obtaining the object image and deleting the object according to the complete object wireframe. The corresponding object in the image is removed to obtain an object parameter image, the object parameter image is identified as an object parameter, and the object parameter and its coordinates are aggregated into an object parameter information; an arrow start and end point detection step is performed to obtain the object image and delete the portion of the image corresponding to the object image and the object parameter image to obtain an arrow image, and then the start and end points of each arrow in the arrow image and the coordinates of the start and end points of the arrows are identified and aggregated into an arrow. The method comprises the steps of: determining the identification results of the frame information identification step, the object detection step, the object parameter detection step and the arrow start and end point detection step, corresponding the object parameter coordinates to the arrow start and end point coordinates, and then corresponding the arrow start and end point coordinates to the object coordinates. According to the determination result, the object and the object parameter are matched, and the matching result is output as an object parameter matching information.

In one embodiment, an image pre-processing step is further included, which uses an image processing algorithm to remove line segments, text and noise from the object image, obtain an object frame and at least one sub-object frame thereof, and merge the object frame and the sub-object frame to obtain the complete object wireframe of the object.

In one embodiment, the object detection step is to read the object image, obtain the outermost object frame image based on the complete object wireframe image, scan the outermost object frame image to obtain an object width, segment each object based on the object width and the object-related information, and then summarize the object name of each object and the corresponding object coordinates to obtain an object information.

In one embodiment, the object parameter detection step is to delete the object in the object image corresponding to the complete object wireframe to obtain the object parameter graph, then identify the object parameter graph as the object parameter, and aggregate the object parameter and its corresponding object parameter coordinates into object parameter information.

In one embodiment, the arrow start and end point detection step is to delete the object in the object image that corresponds to the complete object wireframe diagram, and then remove the object parameters corresponding to the object parameter diagram to obtain an arrow diagram, and perform a skeletonization algorithm on the arrow diagram to obtain each of the arrow endpoints, and identify and aggregate the coordinates of each of the arrow start and end points based on the density of each of the arrow endpoints to output.

In one embodiment, a production output step is further included, which is to output the corresponding information of the object parameters to a computer numerical control processing equipment to produce each of the drawing objects.

The directions or similar terms described throughout the present invention, such as front, back, left, right, top, bottom, inside, outside, side, etc., are mainly with reference to the directions of the drawings. Each direction or similar terms are only used to assist in explaining and understanding the various embodiments of the present invention, and are not used to limit the present invention.

The articles used in the elements and components described in the present invention, such as a or the, are only for the convenience of use or simplified description and should be interpreted as including one or at least one, and a single concept also includes a plural concept unless it is obvious that there is a different meaning.

In order to make the above and other purposes, features and advantages of this invention more clearly understood, the following is a preferred embodiment of this invention and a detailed description thereof with accompanying drawings. The same symbols in different drawings are considered the same and their descriptions are omitted.

FIG1 is a system block diagram of an embodiment of the present invention, and FIG2 is a system block diagram of a text and graphic element recognition system of another embodiment of the present invention. Referring to FIG1 and FIG2, the text and graphic element recognition system of the present embodiment is capable of processing a pixelated drawing 10 of a bitmap-type engineering drawing, and the drawing 10 includes a frame area 11 and a graphic area 12. The so-called frame area 11 includes the drawing frame, parts list and other contents, and the so-called graphic area 12 includes the drawing contents outside the frame area 11. The text and graphic element recognition system of the drawing includes: a frame information recognition unit 20, an object detection unit 30, an object parameter detection unit 40, an arrow start and end point detection unit 50 and a detection parameter attribution unit 60. The frame information recognition unit 20 can recognize the text image of the frame area 11 in the drawing 10 as a frame information 22. The frame information 22 is in character format and can be output as a data exchangeable frame information file, such as a Json file, as needed to facilitate exchange between different devices or systems. The frame information file can include field types and field values such as object name, specification, total length, shoulder, quantity, material, etc.

The frame information recognition unit 20 captures a frame image 21 from the preset coordinates of the frame area 11, and recognizes the frame image 21 as a frame information 22 in a character format. The frame information 22 includes information such as the total length of an object and the type and quantity of shoulders.

The object detection unit 30 scans the graphic area 12 to capture an object image 31, performs image processing on the object image 31 to obtain a complete object wireframe 32 and an outermost object border image 321, scans the outermost object border image 321 to obtain an object width, and divides at least one object 33 according to the object width and the total length of the object and the type and number of the shoulders in the frame information 22, and integrates an object name 34 and an object coordinate 35 of each object 33 into an object information 36.

The object parameter detection unit 40 obtains the object image 31 of the graphic area 12 and deletes the object 33 corresponding to the complete object wireframe 32 to obtain an object parameter diagram 41, identifies the text graphic in the object parameter diagram 41 as an object parameter 42 in character format and an object parameter coordinate 43 thereof, and aggregates the text into object parameter information 44.

The arrow start and end point detection unit 50 obtains the object image 31 of the graphic area 12 and deletes the object 33 corresponding to the complete object wireframe image 32 and the object parameter 42 corresponding to the object parameter image 41 to obtain a plurality of arrow images 51, and then uses image processing technology to detect each arrow start and end point 52 and a corresponding arrow start and end point coordinate 53 in each arrow image 51, and aggregates them into arrow start and end point information 54.

The detection parameter attribution unit 60 reads the frame information 22, the object information 36, the object parameter information 44 and the arrow start and end point information 54, corresponds the object parameter coordinates 43 and the arrow start and end point coordinates 53, and then corresponds each of the arrow start and end point coordinates 53 and the object coordinates 35, determines the relationship between each of the objects 33 and each of the object parameters 42, and summarizes and outputs an object parameter corresponding information 61 in the form of a text file.

Similarly, the detection and aggregation results of the frame information recognition unit 20, the object detection unit 30, the object parameter detection unit 40, the arrow start and end point detection unit 50 and the detection parameter attribution unit 60 can also be output as an information file for data exchange as needed.

As shown in FIG. 2 , the present embodiment may further include a production output unit 70 : that is, the object parameter corresponding information 61 is introduced into a computer numerical control processing equipment, so as to use the computer numerical control processing equipment to manufacture the physical product of the drawing 10 .

FIG3 is a method flow chart of a method for recognizing text and graphics elements in a drawing of an embodiment of the present invention. Referring to FIG3, the method for recognizing text and graphics elements in a drawing of the present embodiment can process an inputted raster-type engineering drawing 10, wherein the drawing 10 includes a frame area 11 and a graphic area 12, and the portion outside the frame area 11 can be regarded as the graphic area 12. The method for recognizing text and graphics elements in a drawing includes: a frame information recognition step S20, an object detection step S30, an object parameter detection step S40, an arrow start and end point detection step S50, and a detection parameter attribution step S60.

The frame information recognition step S20 captures a frame image 21 according to the frame area 11, divides the text intervals of the frame image 21 into a plurality of single text images, and recognizes each of the single text images using a pre-trained deep learning model to perform recognition and reorganization of each single character and aggregate them into a frame information 22.

The object detection step S30 captures an object image 31 from the graphic area 12 of the drawing 10, identifies a complete object wireframe 32 and an outermost object border image 321 in the object image 31 using an image processing algorithm, reads and scans the outermost object border image 321 to segment at least one object 33 based on the frame information 22, and aggregates each object name 34 and an object coordinate 35 thereof into an object information 36.

The object parameter detection step S40 obtains the object image 31 from the drawing 10, and deletes the corresponding object 33 in the object image 31 according to the complete object wireframe 321 to obtain an object parameter diagram 41, identifies the object parameter diagram 41 as an object parameter 42, and aggregates the object parameter 42 and an object parameter coordinate 43 thereof into an object parameter information 44.

The arrow start and end point detection step S50 obtains the object image 31 from the drawing 10, and deletes the object 33 corresponding to the complete object wireframe diagram 321 and the object parameter 42 corresponding to the object parameter diagram 41 in the object image 31 to obtain an arrow diagram 51, and then identifies each arrow start and end point 52 of the arrow diagram 51 and its corresponding arrow start and end point coordinates 53, and aggregates them into arrow start and end point information 54.

The detection parameter belongs to step S60, which uses an image processing algorithm to determine the identification and detection results of the frame information identification step S20, the object detection step S30, the object parameter detection step S40 and the arrow start and end point detection step S50, and corresponds the object parameter coordinates 43 to the arrow start and end point coordinates 53, and then corresponds the arrow start and end point coordinates 53 to the object coordinates 35. According to the above determination results, the object 33 and the object parameter 42 are matched, and the matching result is output as an object parameter matching information 61.

FIG4 is a method flow chart of a method for recognizing text and graphic elements in a drawing of another embodiment of the present invention. Referring to FIG4, the present embodiment further includes an image pre-processing step S10 for the complete object wireframe 32. The image pre-processing step S10 uses an image processing algorithm to remove graphic elements such as marked line segments, annotation text, and miscellaneous dots in the object image 31 to obtain an object frame and its sub-object frame, and then merges the object frame and the sub-object frame to obtain the complete object wireframe 32 of the object 33.

In detail, the object detection step S30 reads the object image 31, obtains the outermost object frame image 321 based on the complete object wireframe image 32, scans the outermost object frame image 321 to obtain an object width, and divides each object 33 according to the object width and the object-related information in the frame information, such as the total length of the object and the number of shoulders, and then summarizes the object name 34 of each object 33 and the corresponding object coordinates 35 to obtain an object information 36.

The object parameter detection step S40 is to delete the object 33 in the object image 31 corresponding to the complete object wireframe 32 to obtain the object parameter diagram 41, then identify the object parameter diagram 41 as the object parameter 42, and integrate the object parameter 42 and the corresponding object parameter coordinates 43 into an object parameter information 44.

The arrow start and end point detection step S50 obtains the object image 31 from the drawing 10, deletes each object 33 corresponding to the complete object wireframe 32 in the object image 31, and then removes each object parameter 42 corresponding to the object parameter image 41 in the object image 31 to obtain an arrow image 51, performs a skeletonization algorithm operation on the arrow image 51 to obtain each arrow end point, identifies each arrow start and end point 52 according to the density of each arrow end point, and summarizes each arrow start and end point coordinate 53 corresponding to it as an arrow start and end point information 54.

It is worth mentioning that, as shown in FIG. 4 , the text and graphic element recognition method of the drawing of the present invention may further include a production output step S70 for outputting the object parameter corresponding information 61 to a computer numerical control processing equipment to automatically produce the objects 33 of each drawing 10.

On the other hand, the frame information 22 of the frame information identification step S20, the object information 36 of the object detection step S30, the object parameter information 44 of the object parameter detection step S40, the arrow start and end point information 54 of the arrow start and end point detection step S50, and the object parameter corresponding information 61 of the detection parameter attribution step S60 can be output as a text file that can be exchanged for data to be exchanged with other devices, thereby expanding the scope of application.

Figures 5 to 15 are schematic diagrams of the drawings 10 and the text information summarized in each step corresponding to each step of Figure 3. Figure 5 shows that the present invention can process the contents of a drawing 10 of a bitmap engineering drawing, and the drawing 10 can include various graphic elements such as frame information 22, parts list information, objects 33, object parameters 42, etc.; the box in Figure 6 indicates the location of the frame information 22; in Figure 7, the double quotation marks before the colon are the name of the frame information 22, and the double quotation marks after the colon are its value. FIG8 is a schematic diagram of a complete object wireframe diagram of the object detection step in the drawing of the first embodiment of the present invention. The complete object wireframe diagram 32 is a stage result after deleting the annotation line segments, annotation text, noise and object border noise from the acquired object image 31. FIG9 is a schematic diagram of an object information segment output by the object detection step of the first embodiment of the present invention. In the figure, the object information segment 36 includes an object name 34 and object coordinates 35. FIG10 is a schematic diagram of an object parameter diagram of the object parameter detection step in the drawing of the first embodiment of the present invention. The object parameter diagram 41 is a stage result after deleting the object 33 from the acquired object image 31. FIG. 11 is a schematic diagram of a fragment of object parameter information summarized in the object parameter detection step of the first embodiment of the present invention, wherein the fragment of object parameter information 44 is shown, which includes object parameters 42 and object parameter coordinates 43. FIG. 12 is a schematic diagram of an arrow diagram of the arrow start and end point detection step in the drawing of the first embodiment of the present invention, wherein the arrow diagram 51 is a stage result after deleting the corresponding object 33 and the corresponding object parameter 34 from the obtained graphic area 12, and wherein the positions of the arrow start points 521 of the arrow start and end points 52 are marked with dots, and the positions of the arrow end points 522 are marked with squares. FIG. 13 is a schematic diagram of a fragment of arrow start and end point information summarized in the arrow start and end point detection step of an embodiment of the present invention, wherein the fragment is a fragment of arrow start and end point information 54, which includes the arrow number and its arrow start and end point coordinates 53. FIG. 14 is a schematic diagram of objects and corresponding object parameters in the detection parameter attribution step in the drawing of the embodiment of the present invention, wherein different groups but corresponding objects 33 and object parameters 42, and objects 33' and object parameters 42' are marked. FIG. 15 is a schematic diagram of object parameter corresponding information summarized in the detection parameter attribution step of the embodiment of the present invention, wherein the text fragment of object parameter corresponding information 61 is shown.

In summary, this invention can recognize drawings in paper or image file formats, and extract the parameters including object specifications, dimensions, processing tolerances, etc. in the input drawing into a text file through a frame information recognition step, an object detection step, an object parameter detection step, an arrow start and end point detection step, and a detection parameter attribution step. The text file can be directly input into a computer numerical control processing equipment to perform object production operations.

Although the present invention has been disclosed by using the above preferred embodiments, they are not intended to limit the present invention. Any person with ordinary knowledge in the field can make various changes and modifications to the above embodiments within the spirit and scope of the present invention, and they are still within the technical scope protected by the present invention. Therefore, the protection scope of the present invention includes all changes within the meaning and equivalent scope recorded in the attached patent application scope. In addition, when the above several embodiments can be combined, the present invention includes the implementation of any combination.

(10): Drawing (11): Frame area (12): Graphic area (20): Frame information identification unit (21): Frame image (22): Frame information (30): Object detection unit (31): Object image (32): Complete object wireframe (321): Outermost object frame (33,33’): Object (34): Object name (35): Object coordinates (36): Object information (40): Object parameter detection unit (41): Object parameter diagram 42,42’): Object parameters (43): Object parameter coordinates (44): Object parameter information (50): Arrow start and end point detection unit (51): Arrow diagram (52): Arrow start and end points (521): Arrow start point (522): Arrow end point (53): Arrow start and end point coordinates (54): Arrow start and end point information (60): Detection parameter belonging unit (61): Object parameter corresponding information (70): Production output unit S10~S70: Steps

The implementation of this creation will be described in the following simple explanation with drawings. Figure 1 is a system block diagram of a text and graphic element recognition system for a drawing of an embodiment of this creation. Figure 2 is a system block diagram of a text and graphic element recognition system for a drawing of another embodiment of this creation. Figure 3 is a method flow chart of a method for text and graphic element recognition for a drawing of an embodiment of this creation. Figure 4 is a method flow chart of a method for text and graphic element recognition for a drawing of another embodiment of this creation. Figure 5 is a diagram of a processable dot matrix engineering drawing of an embodiment of this creation. Figure 6 is a diagram of a frame picture of a frame information recognition step in a drawing of an embodiment of this creation. Figure 7 is a schematic diagram of the frame information collected in the frame information recognition step of the first embodiment of the present invention. Figure 8 is a schematic diagram of the complete object wireframe diagram of the object detection step in the drawing of the first embodiment of the present invention. Figure 9 is a schematic diagram of the object information fragment output by the object detection step of the first embodiment of the present invention. Figure 10 is a schematic diagram of the object parameter diagram of the object parameter detection step in the drawing of the first embodiment of the present invention. Figure 11 is a schematic diagram of the object parameter information fragment collected in the object parameter detection step of the first embodiment of the present invention. Figure 12 is a schematic diagram of the arrow diagram of the arrow start and end point detection step in the drawing of the first embodiment of the present invention. FIG. 13 is a schematic diagram of the arrow start and end point information fragments summarized in the arrow start and end point detection step of the first embodiment of the present invention. FIG. 14 is a schematic diagram of the objects and corresponding object parameters in the detection parameter attribution step in the drawing of the present invention. FIG. 15 is a schematic diagram of the object parameter corresponding information fragments summarized in the detection parameter attribution step of the present invention.

(10): Drawing

(11): Frame area

(12): Graphic area

(20): Frame information recognition unit

(21):Frame image

(22):Frame information

(30): Object detection unit

(31):Object image

(32): Complete object wireframe

(321): Outermost object border image

(33): Objects

(34): Object name

(35): Object coordinates

(36): Object information

(40): Object parameter detection unit

(41): Object parameter diagram

(42): Object parameters

(43): Object parameter coordinates

(44): Object parameter information

(50): Arrow start and end point detection unit

(51): Arrow diagram

(52): Arrow start and end points

(53): Arrow start and end point coordinates

(54): Arrow start and end point information

(60): Detection parameter belonging to unit

(61): Object parameter corresponding information

Claims (9)

A drawing text and graphic element recognition system is suitable for processing a drawing (10) of a bitmap engineering drawing. The drawing (10) includes a frame area (11) and a graphic area (12). The drawing text and graphic element recognition system includes: a frame information recognition unit (20) for capturing a frame image (21) from the frame area (11) according to a preset coordinate in the drawing (10) and recognizing frame information (22) of the frame image (21); an object detection unit A unit (30) captures an object image (31) according to the graphic area (12) in the drawing (10), detects an object width of the object image (31), segments at least one object (33) according to the object width and the frame information (22), and aggregates an object name (34) and an object coordinate (35) of each object (33) into object information (36); an object parameter detection unit (40) deletes the object corresponding to the image in the drawing (10). The invention relates to a method for detecting an object parameter map (41) by deleting a portion of the object (33) and the object parameter map (41) to obtain an object parameter map (41), identifying the object parameter map (41) as an object parameter (42), and integrating the object parameter (42) and an object parameter coordinate (43) thereof into an object parameter information (44); an arrow start and end point detection unit (50) deletes a portion corresponding to the object (33) and the object parameter map (41) in the drawing (10) to obtain an arrow map (51), and then detecting the arrow map (51) 1) each arrow start and end point (52) and a corresponding arrow start and end point coordinate (53), and aggregated into arrow start and end point information (54); and a detection parameter attribution unit (60), reads the frame information (22), the object information (36), the object parameter information (44) and the arrow start and end point information (54), uses an image algorithm to determine the relationship between each object (33) and each object parameter (42), and aggregates object parameter corresponding information (61). The text and graphic element recognition system of the drawing as described in claim 1 further comprises a production output unit (70) for inputting the object parameter corresponding information (61) into a computer numerical control processing device that is informationally connected to the detection parameter attribution unit (60) to produce a physical product of the drawing (10). A method for recognizing text and graphic elements in a drawing is suitable for processing an inputted drawing of a raster-type engineering drawing. The drawing includes a frame area and a graphic area. The method for recognizing text and graphic elements in a drawing includes: a frame information recognition step (S20), capturing a frame image according to the frame area, and recognizing the text pattern of the frame image as frame information by using an image processing algorithm; an object detection step (S30), capturing an object image according to the graphic area; An object image is obtained and a complete object wireframe and an outermost object frame are identified by an image processing algorithm. The outermost object frame is read and scanned according to the frame information to segment at least one object, and an object name and an object coordinate of each object are collected as object information; an object parameter detection step (S40) is performed to obtain the object image and delete the corresponding object in the image according to the complete object wireframe to obtain an object parameter. a detection step (S50) of obtaining the object image and deleting the portion of the image corresponding to the object image and the object parameter image to obtain an arrow image, and then identifying the start and end points of each arrow in the arrow image and the coordinates of the start and end points of the arrows and the coordinates of the start and end points of the arrows to be aggregated into the start and end point information of the arrows; and a detection parameter attribution step (S51). 60), determine the identification and detection results of the frame information identification step (S20), the object detection step (S30), the object parameter detection step (S40) and the arrow start and end point detection step (S50), correspond the object parameter coordinates with the arrow start and end point coordinates, and then correspond the arrow start and end point coordinates with the object coordinates, and correspond the object and object parameter according to the determination result, and summarize the correspondence result into object parameter correspondence information. The method for recognizing text and graphic elements in a drawing as described in claim 3 further includes an image pre-processing step (S10), which uses an image processing algorithm to remove line segments, text and noise from the object image, obtain an object bounding box and at least one sub-object bounding box, and merge the object bounding box and the sub-object bounding box to obtain the complete object wireframe of the object. The method for recognizing text and graphic elements of a drawing as described in claim 4, wherein the object detection step (S30) is to read the object image, obtain the outermost object frame diagram based on the complete object wireframe diagram, scan the outermost object frame diagram to obtain an object width, segment each object based on the object width and the object-related information, and then summarize the object name of each object and the corresponding object coordinates to obtain an object information. The method for identifying text and graphic elements in a drawing as described in claim 5, wherein the object parameter detection step (S40) is to delete the object corresponding to the complete object wireframe in the object image to obtain the object parameter diagram, then identify the object parameter diagram as the object parameter, and aggregate the object parameter and the corresponding object parameter coordinates into an object parameter information. The method for identifying text and graphic elements in a drawing as described in claim 6, wherein the arrow start and end point detection step (S50) is to delete the object corresponding to the complete object wireframe in the object image, and then remove the object parameter corresponding to the object parameter image to obtain an arrow image, perform a skeletonization algorithm on the arrow image to obtain each arrow end point, identify each arrow start and end point according to the density of each arrow end point, and summarize each arrow start and end point and its corresponding coordinates. The method for recognizing text and graphic elements in a drawing as described in claim 3 further includes a production output step (S70), which is to output the corresponding information of the object parameters to a computer numerical control processing equipment to produce each object in the drawing. The method for identifying text and graphic elements in a drawing as described in claim 3 further includes outputting the frame information, the object information, the object parameter information, the arrow start and end point information, and the object parameter corresponding information into a text file that can be exchanged.