WO2015076343A1 - Data-stitching device, data-stitching method, and computer program - Google Patents

Abstract

This invention minimizes inconsistencies in connecting regions when performing data stitching. This data-stitching device (1) for stitching together a plurality of pieces of data each containing a plurality of data elements has a processing unit (11) that considers each piece of data to be shape data for an elastic body and determines elastic deformation that occurs in the respective elastic bodies corresponding to the plurality of pieces of data. The shape of each elastic body is the shape that connects values represented by the data elements constituting that piece of data. The aforementioned elastic deformation is that which occurs when the plurality of elastic bodies are connected to each other such that positions thereon that correspond to data elements whose values should coincide across multiple pieces of data coincide. The processing unit (11) outputs, as stitched data representing the plurality of pieces of data stitched together, data representing the shapes of the elastically deformed elastic bodies.

Description

Data stitching apparatus, data stitching method, and computer program

The present invention relates to an apparatus, a method, and a computer program for stitching a plurality of data.

When measuring an object that exceeds the measurement range of the measurement device, it is necessary to stitch (connect) independent measurement data.
For example, Patent Documents 1 and 2 disclose a method of calculating shape data of the entire test surface by connecting shape data obtained by partially measuring the test surface.
Data stitching (data connection) is not limited to stitching of shape measurement data, but is performed on various data such as stitching of a plurality of image data.

International Publication No. 2011/061843 International Publication No. 2012/008031

Since individual measurement data to be stitched has an error independently, it is necessary to obtain a plausible value (maximum likelihood value) at the time of data stitching.
Generally, in the conventional data stitching, intermediate point values are generated so that the value of the connection area in each data is likely. Then, using the intermediate point as a bridge, two different data are connected.

単 純 Simple average is used to calculate the midpoint. As a relatively advanced method, there is a method of adding rotation and offset values (a total of six parameters including three parameters related to translation and three parameters related to rotation) to all data in each data group. The least square method is often used as a method for determining these parameters.

However, even when stitching (connection) such as three-dimensional data is performed using only the above parameters, the discontinuity of values in the data connection area is not eliminated. In addition, the connection error is distributed only to the data connection region, and tends to be a value far from a likely value (maximum likelihood value).

Therefore, an object of the present invention is to provide a new technical means for suppressing a mismatch of connection areas in data stitching.

(1) The present invention from a certain viewpoint is a data stitching device for stitching a plurality of data each having a plurality of data elements,
Each of the plurality of data is regarded as data of the shape of the elastic body, and includes a processing unit that obtains elastic deformation generated in each of the plurality of elastic bodies corresponding to the plurality of data,
The shape of the elastic body is a shape connecting values indicated by a plurality of data elements included in the data,
The elastic deformation is caused by connecting the plurality of elastic bodies to each other so that the positions coincide with each other at the position of the elastic body corresponding to the data element that should have the same data element value in the plurality of data. And
The processing unit is a data stitching device that outputs data indicating the shapes of the plurality of elastic bodies having undergone the elastic deformation as stitching data obtained by stitching the plurality of data. In addition to the case where the relative distance between the elastic body parts to be connected becomes completely zero, the case where the relative distance becomes substantially zero, that is, the case where the relative distance is such that it can be regarded as substantially zero is included.

The present invention performs data stitching by an approach completely different from conventional data stitching, in which analysis of elastic deformation is applied to data analysis called data stitching.

(2) It is preferable that the processing unit is configured to set an elastic modulus of each of the plurality of elastic bodies.

(3) It is preferable that the said process part can set a different elastic modulus to each of a some elastic body.

(4) It is preferable that the processing unit is configured to determine an elastic modulus of each of the plurality of elastic bodies based on reliability of each of the plurality of data.

(5) The processing unit is configured to perform a process of adding an interpolation data element for interpolating the plurality of data elements to at least one of the plurality of data, and the value of the data element in the plurality of data The data elements that should be matched preferably include the interpolated data elements.

(6) The processing unit is configured to obtain internal stress of the plurality of elastic bodies in which the elastic deformation has occurred, and detect inappropriate data among the plurality of data based on the internal stress. Is preferred.

(7) It is preferable that the processing unit obtains internal stress of the plurality of elastic bodies in which the elastic deformation has occurred, and detects inappropriate connection of the plurality of elastic bodies based on the internal stress.

(8) It is preferable that the processing unit obtains the elastic deformation by deformation analysis using a plurality of elastic body deformation models generated based on the plurality of data.

(9) It is preferable that each of the plurality of elastic body deformation models includes at least a plurality of elements that can be bent and deformed.

(10) The element preferably includes a beam element.

(11) The element preferably includes a shell element.

(12) The deformation analysis is preferably performed by a finite element method.

(13) It is preferable that the elastic deformation includes at least bending deformation.

(14) The plurality of data are a plurality of shape measurement data obtained by measuring the surface of the measurement object along a plurality of measurement paths,
Each of the plurality of data elements is a data element indicating a measurement value on the measurement path,
Each of the plurality of measurement paths has an intersection with another measurement path included in the plurality of measurement paths,
The shape of the elastic body is a shape of a line connecting the measurement values indicated by a plurality of data elements included in the shape measurement data,
The elastic deformation is preferably elastic deformation generated by connecting the plurality of elastic bodies so that the positions coincide with each other at the position of the elastic body corresponding to the data element indicating the measurement value of the intersection.

(15) The plurality of data is image data,
Each of the plurality of data elements is a pixel;
Each of the shooting ranges of the plurality of image data has an overlapping area with shooting ranges of other image data included in the plurality of image data,
The shape of the elastic body is a shape of a surface connecting pixel values indicated by a plurality of pixels included in the image data,
The elastic deformation is preferably elastic deformation generated by connecting the plurality of elastic bodies so that the positions coincide with each other at the positions of the elastic bodies corresponding to the pixels included in the overlapping region.

(16) The plurality of data are a plurality of observation data obtained by observing an observation object with an observation device,
Each of the observation ranges of the plurality of observation data preferably has an overlapping area with the observation ranges of other observation data included in the plurality of observation data.

(17) The processing unit is configured to perform a process of evaluating how to observe the observation object,
The processing to be evaluated is preferably performed based on a result of eigenvalue analysis on the whole of the plurality of elastic bodies in which the elastic deformation has occurred.

(18) The present invention from another viewpoint is a method in which a computer performs a process of stitching a plurality of data each having a plurality of data elements,
Each of the plurality of data is regarded as data of the shape of the elastic body, and the computer executes a process for obtaining elastic deformation occurring in each of the plurality of elastic bodies corresponding to the plurality of data,
The computer outputs data indicating the shapes of the plurality of elastic bodies in which the elastic deformation has occurred as stitching data obtained by stitching the plurality of data,
The shape of the elastic body includes a value connecting values indicated by a plurality of data elements included in the data,
The elastic deformation is caused by connecting the plurality of elastic bodies to each other so that the positions coincide with each other at the position of the elastic body corresponding to the data element that should have the same data element value in the plurality of data. It is.

(19) The present invention from another point of view
A computer program for causing a computer to execute a process of stitching a plurality of data each having a plurality of data elements,
The processing is as follows:
Each of the plurality of data is regarded as data of the shape of the elastic body, obtaining elastic deformation generated in each of the plurality of elastic bodies corresponding to the plurality of data;
Outputting data indicating the shapes of the plurality of elastic bodies in which the elastic deformation has occurred, as stitching data obtained by stitching the plurality of data;
Including
The shape of the elastic body is a shape connecting values indicated by a plurality of data elements included in the data,
The elastic deformation is caused by connecting the plurality of elastic bodies to each other so that the positions coincide with each other at the position of the elastic body corresponding to the data element that should have the same data element value in the plurality of data. It is.
From another viewpoint, the present invention is a recording medium on which the computer program is recorded.

According to the present invention, in data stitching, it is possible to suppress inconsistency of connection areas and output a result closer to the true value.

It is a block diagram of a data stitching apparatus. It is a top view which shows the measurement path | pass in a measurement object. It is explanatory drawing of measurement data. This is measurement data before stitching processing. It is a flowchart of a stitching process. It is explanatory drawing of the element division | segmentation of a linear elastic body. It is a figure which shows the elastic body deformation | transformation model of a planar elastic body. It is explanatory drawing of a data point setting. It is the measurement data after stitching processing. This is measurement data before stitching processing. It is the measurement data after stitching processing. It is a flowchart of an evaluation process. It is explanatory drawing of a measurement path | pass eigenvalue. It is explanatory drawing of a measurement path | pass and an eigenvalue. It is explanatory drawing of an imaging | photography range and an eigenvalue. It is explanatory drawing of data and an elastic body shape. It is explanatory drawing of an elastic body connection. It is explanatory drawing of an elastic body connection and elastic deformation. It is explanatory drawing of a displacement function. It is data explanatory drawing before and after stitching.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

FIG. 1 shows a stitching apparatus 1 according to the embodiment. The stitching processing apparatus 1 described below is used for stitching (connecting) a plurality of data (data to be stitched) to obtain stitching data. The stitching device 1 according to the embodiment handles a plurality of measurement data output from the measurement device 2 as a plurality of stitched data. A wide range of measurement data can be obtained by performing data stitching.

The “stitched data” is not limited to the measurement data, and for example, the data (for example, measurement data) obtained for a part of a specific target (for example, measurement target or imaging target). Or image data).

The stitching device 1 according to the present embodiment includes a data acquisition unit 2 and a processing device 3. The data acquisition unit 2 in the present embodiment is configured as a measurement device that measures the surface shape of a (measurement) object and outputs measurement data to the processing device 3. However, the data acquisition unit 2 is not limited to the measurement device, and may be a camera that outputs image data to be stitched data, for example. That is, the data acquisition part 2 should just be what can acquire the data about a specific object. In other words, the data acquisition unit 2 only needs to be able to observe an observation target (for example, a measurement target or a photographing target) with an observation device (for example, a measurement device or a camera) and acquire observation data of the observation target. Observation devices such as measurement devices and cameras have a limited observation range (for example, a measurement range and an imaging range). A limited observation range can be expanded by data connection (data stitching) of a plurality of observation data having an overlap region of the observation range in the overlap region.
Further, the data acquisition unit 2 may not be provided, and the processing device 3 acquires and processes the stitched data stored in the storage unit 12 in the processing device 3 or an external storage device. Also good.

The processing device 3 is configured by a computer, and includes a processing unit 11, a storage unit 12, an input unit 13 such as a keyboard / mouse, and an output unit 14 such as a display.
The processing unit 11 exhibits the functions of the processing device 3 by executing the computer program stored in the storage unit 12. Specifically, as illustrated in FIG. 1, the processing unit 11 includes a data acquisition processing unit 11a, an elastic modulus setting unit (deformation model generation unit) 11b, a coincidence data point setting unit 11c, a deformation calculation unit 11d, and a detection unit 11e. The functions of the output processing unit 11f and the evaluation unit 11g are exhibited. Details of each functional unit will be described later.

The measurement apparatus (data acquisition unit) 2 of the present embodiment performs a one-dimensional scan along the measurement path on the measurement target surface (xy plane) in order to measure the fine uneven shape of the measurement target surface. Thus, the thickness direction position (height; z direction position) of the measurement target is obtained.
FIG. 2 shows a plurality of measurement paths when the measuring device 2 measures a circular flat plate.

The measurement path shown in FIG. 2 includes 60 linear measurement paths (straight path) and one annular measurement path (annular path). Each measurement path has an intersection with another measurement path. An intersection of measurement paths is an overlapping range (overlapping point) of measurement ranges along a plurality of measurement paths. The intersection (overlapping range) is the position where data is connected.
The straight paths are arranged at intervals of 6 ° on the circular flat plate, and the individual straight paths intersect with many other straight paths. The annular path is arranged on the outer peripheral side of the circular flat plate, and is an annular path that connects the distal ends on the radially outer side of all straight paths.
In addition, since the center part of the circular flat plate is outside the measurement range, the measurement range is a donut-shaped region.

The measurement device 2 outputs measurement data (stitched data) obtained by measuring the surface shape of the measurement target along each measurement path. For example, the measurement data for the first measurement path (straight line path) and the second measurement path (straight line path) shown in FIG. 3A has a data structure as shown in FIGS. 3B and 3C. Have.

The first measurement path shown in FIG. 3A connects the coordinates (x _1n , y _1n ) from the coordinates (x ₁₁ , y ₁₁ ) on the xy plane that are parallel to the surface of the circular flat plate to be measured. It is a straight path. As shown in FIG. 3B, the measurement data for the first measurement path includes a plurality of (n) measurement points (sampling points) from the coordinates (x ₁₁ , y ₁₁ ) to the coordinates (x _1n , y _1n ). Each of the measurement target thickness direction positions (height: z direction position) is configured as a data string having data results as measurement results.
The second measurement path shown in FIG. 3A is a straight path connecting the coordinates (x _2n , y _2n ) to the coordinates (x ₂₁ , y ₂₁ ) on the xy plane. As shown in FIG. 3C, the measurement data for the second measurement path includes a plurality (n) of measurement points (sampling points) from the coordinates (x ₂₁ , y ₂₁ ) to the coordinates (x _2n , y _2n ). Each of the measurement target thickness direction positions (height: z direction position) is configured as a data string having data results as measurement results.
Thus, the measurement data is a data string of three-dimensional data in which each data element has a three-dimensional value of x, y, z.

Individual measurement data is only a measurement result of the measurement range in a part of the measurement path when viewed from the circular flat plate that is the measurement target, so to obtain the measurement result of the entire measurement object, along the multiple measurement paths It is necessary to stitch (connect) a plurality of measurement data obtained by measurement at intersections (overlap ranges) of measurement paths.

However, each measurement data has an independent error, and the values of a plurality of data to be connected do not always match even at a point (intersection) to which the data is to be connected. For example, in FIG. 3, since the same position of the measurement object is measured at the intersection (coordinates (x _a , y _a )) of the first measurement path and the second measurement path, the true values are naturally the same. However, each measurement data has an error such as a measurement error independently. For this reason, even at the intersection (coordinates (x _a , y _a )) that is the overlapping range of the first measurement path and the second measurement path, the measurement data values (z-direction positions) of the first measurement path and the second measurement path are the same. Usually they do not match.

FIG. 4 illustrates measurement data (data before processing by the processing device 3) measured by the measuring device 2 along all the measurement paths shown in FIG. 2 on the x, y, and z coordinates. As shown in FIG. 4, due to errors included in individual measurement data, the measurement data of the straight path does not always coincide with the measurement data of the other straight paths even at the position that should be the intersection. In addition, the measurement data of the circular path does not always coincide with the radially outer tip of the measurement data of the straight path.
Thus, each of the plurality of measurement data is mutually contradictory data due to an error.

Therefore, when data stitching is performed, it is necessary to eliminate the mismatch of the values at the original intersection and to obtain a plausible data. In general, using the fact that the true value of the intersection point is common, the average point of the values of multiple data elements corresponding to each intersection point is obtained, and each average point is interpolated with a polygonal plane, etc. It is conceivable to take this method.
However, in this case, the mismatch of values is only resolved in the vicinity of the intersection, and as a whole, the inconsistency of the data values remains around the intersection. That is, inconsistency in the data connection area remains.

Therefore, the processing device 3 of the present embodiment regards the measurement data group as an elastic body data group. That is, each measurement data (stitched data) is treated as a model of an object (elastic body) having a shape defined by the value of the measurement data. In the case of measurement data, the shape of the elastic body is a line shape that connects values indicated by the data elements constituting the measurement data in three-dimensional coordinates. If the shape connecting the values indicated by the data elements constituting the stitched data becomes a surface, the shape of the elastic body becomes a surface shape. If the shape connecting the values indicating the data elements constituting the stitched data becomes a three-dimensional shape, the shape of the elastic body becomes a three-dimensional shape.
By regarding the measurement data as an elastic body, the data stitching (data connection) problem is replaced with an elastic body connection problem. In the present embodiment, the inconsistency of the data connection area is solved by solving the elastic body connection problem.

FIG. 5A shows the procedure of the stitching process for performing data stitching by solving the elastic body connection problem.
First, the data acquisition processing unit 11 a of the processing unit 11 of the processing device 11 performs a data acquisition process for controlling the measurement of the data acquisition unit 2. The data acquisition processing unit 11a receives a plurality of measurement data (stitched data) corresponding to a plurality of measurement paths in the same measurement target from the measurement device 2.

Then, the elastic modulus setting unit (physical property value setting unit) 11b of the processing unit 11 of the processing device 3 sets an elastic modulus (such as Young's modulus) for each of a plurality of measurement data (physical property value setting process; step) S1) is performed. In the setting process in step S1, physical property values other than the elastic modulus may be set as necessary. The elastic modulus and other physical property values are used for deformation calculation processing S3 described later. When a physical property value such as an elastic modulus is set in the measurement data, the elastic modulus is set for an elastic body whose shape is indicated by the measurement data. By setting an elastic modulus for each measurement data, the measurement data can be handled as shape data of an elastic body having the set elastic modulus.
Since the measurement data is three-dimensional data as described above, it can be handled as three-dimensional elastic body shape data by setting the elastic modulus. Specifically, a plurality of measurement data is handled as an elastic body set made up of a large number of linear elastic bodies. Each of a large number of linear elastic bodies included in the elastic body set individually exists in the form shown in FIG. 4 in the xyz space.
Although the thickness of the linear elastic body is numerically given, it can be virtually regarded as zero in order to stitch data.
Moreover, the elasticity modulus setting process should just be performed before the start of the below-mentioned deformation | transformation calculation process (step S3), and may be performed after the below-mentioned matching data point setting process (step S2).

The elastic modulus setting unit 11a can also set a common elastic modulus for a plurality of measurement data with respect to a plurality of measurement data (data to be stitched), or a different elastic modulus for each of the plurality of measurement data. Can also be set.
In the present embodiment, since the magnitude of the absolute value of the elastic modulus is not a problem, the magnitude of the elastic modulus may be set as appropriate. When different elastic moduli are set for each of the plurality of measurement data, the elastic modulus set for each of the plurality of measurement data is determined according to the accuracy of the measurement data (data reliability; the magnitude of the included error) (details) Will be described later).

The matching data point setting unit (matching data element setting unit) 11c of the processing unit 11 matches data elements (data points) whose data values (x value, y value, z value) should match among a plurality of measurement data. A matching data point setting process (step S2) to be set is performed as a data point. The coincidence data point setting unit 11c of the present embodiment sets intersections (overlapping areas) of a plurality of measurement paths as coincidence data points. In the case of the measurement path as shown in FIG. 2, the position (xy coordinate) that is the intersection of the measurement path is known in advance, so that the intersection coordinates of the first and second measurement paths are (x _a ) as shown in FIG. , if _{y a),} xy coordinates of the measurement data for each measurement path _(x a, _{y a)} a is a data element _{((x a, y a,} z 1a) and _{(x a, y a,} z _2a )) is set as the coincidence data point, and the measurement data of each measurement path is set. It is preferable that a plurality of matching data points (matching data elements) are set for each measurement data.
The coincidence data points are set corresponding to all the intersections in the measurement path shown in FIG.
The matched data point may be automatically set by the matched data point setting unit 11c, or may be set by the matched data point setting unit 11c based on the designation of the matched data point by the user input.

A plurality of pieces of measurement data in which the elastic modulus is set and the coincidence data point is set are given to the deformation calculation unit 11d. The deformation calculation unit 11d treats each piece of measurement data for which the elastic modulus is set as data of an elastic body having a shape indicated by the value of the measurement data (a shape when no external force is applied). That is, the measurement data is handled as elastic body shape data.
In the deformation calculation unit 11d of the present embodiment, the deformation of the elastic body is obtained by the finite element method. The finite element method is a numerical analysis method generally used in structural analysis and the like, and in this embodiment, the finite element method is used to perform elastic deformation analysis.

The deformation calculation unit (finite element deformation analysis unit) 11d generates an elastic body deformation model (finite element model) obtained by dividing an elastic body having a shape indicated by the measurement data value for deformation analysis. The element division is performed for each of a plurality of elastic bodies corresponding to a plurality of measurement data. The elements constituting the deformation model are, for example, beam elements (one-dimensional elements) or shell elements (two-dimensional elements). Each element constituting the elastic body deformation model is an elastic body element that is at least bendable and has a sectional moment of inertia. Therefore, the elastic deformation model is capable of bending deformation and other elastic deformation. The strain due to the force applied to an arbitrary point of the elastic body deformation model (elastic body) can be transmitted to the entire elastic body deformation model (elastic body).
Beam elements and shell elements are elements used in the finite element method. Therefore, the element division can be performed using a function installed in software of the finite element method.
In the present embodiment, since the elastic body is linear, as shown in FIG. 5B, an elastic body deformation model divided into elements by beam elements is generated. The elastic body deformation model can be configured, for example, by connecting each data element of measurement data by one or a plurality of beam elements.

The stitched data may be image data. The image data is three-dimensional data including two-dimensional position and luminance information, and can be handled in the same manner as measurement data that is three-dimensional data. The image data including the two-dimensional position of each pixel and the luminance value of each pixel can be regarded as elastic body shape data indicating the shape (curved surface shape) of the surface elastic body having irregularities corresponding to the magnitude of the luminance value. In order to generate a deformation model of a planar elastic body, it is preferable to perform element division using a shell element. Even in the case of image data, the data elements may be coupled by beam elements. In this case, the finite element model has a mesh-like surface shape.

FIG. 5C (a) shows curved elastic body deformation models M1 and M2 made of shell elements.
Each image data has an overlapping photographing range (overlapping area). Pixels included in the overlapping area are set as matching data points (matching data elements). A plurality of elastic body deformation models M1, M2 are connected in the overlapping region.

Note that the physical property value (elastic modulus and the like) set in the setting process in step S1 is set as element information for each finite element constituting the elastic body deformation model. Therefore, the elastic body deformation model can be deformed according to the set physical property value (elastic modulus or the like).

The deformation calculation unit 11d matches the measurement data (elastic body shape data; elastic body) whose values may not match each other even in the matching data points, and the values of the data elements match (connection) in all the set matching data points. ). When the values of a plurality of data elements that are matching data points match, the relative values (relative distances) of the plurality of data elements that are matching data points become zero. Thereby, the values of the data elements to be matched can be matched.
With this setting, a plurality of elastic bodies indicated by the elastic deformation model are connected at the position of the coincidence data point. When a plurality of elastic bodies are connected at a position (displacement position) corresponding to the coincidence data point, the position (displacement position) corresponding to the coincidence data point in each elastic body is forcibly displaced. By this forced displacement, each elastic body indicated by the elastic deformation model is elastically deformed.
Due to the forced displacement of the position (displacement position) of the elastic body corresponding to the coincidence data point, bending occurs in each finite element (beam element) constituting the elastic body deformation model, and the elastic deformation due to the forced displacement affects the entire elastic body. To do. As a result, the elastic deformation due to the forced displacement spreads over the connected elastic body aggregates. As a result of such elastic deformation, the deformation calculation unit 11d obtains a shape that the elastic body aggregate naturally takes (a shape that minimizes the sum of energy due to internal stress of the elastic body) by deformation analysis using a finite element method.
The boundary condition of the deformation analysis is performed so that all data points are not over-constrained. That is, a constraint point is arbitrarily selected, and six degrees of freedom (in this embodiment, three translational degrees of freedom and three degrees of freedom of rotation) are constrained. It is also possible to constrain data points with zero error. In this case, over-restraint may be used.

The deformation calculation unit 11d has a constraint that data values of matching data points that should match data values are equal (relative values (relative distances) between displacement positions corresponding to the matching data points in a plurality of elastic bodies become zero. Is given as a constraint for deformation operations. This restriction is physically equivalent to preventing the elastic bodies from being separated at the displacement positions corresponding to the intersections of the measurement paths. Under such restrictions, the deformation calculation unit 11 obtains a shape that the whole connected elastic body aggregate naturally takes.
Note that the setting function for elastic deformation so that arbitrary points of a plurality of elastic bodies are connected is installed in general finite element method software, and the setting can be performed by using such a function. .

Here, the shape of the bamboo basket obtained by weaving bamboo bamboo of various shapes is in a state in which the sum of energy due to the internal stress of bamboo bamboo is minimal. Obtaining the shape of a bamboo basket obtained by weaving bamboo strings is equivalent to obtaining the most likely result when connecting mutually contradictory data (measurement data of this embodiment) at intersections (matching data points). is there.
The shape of the bamboo basket obtained by knitting bamboo can be obtained by the finite element method if the shape and elastic modulus of each bamboo basket (elastic body) are known.
Therefore, the shape of the elastic body set connected by the coincidence data point can be obtained by the finite element method based on the measurement data handled as the elastic body shape data and the elastic modulus in the same manner as the shape of the bamboo basket. it can.

Data indicating the shape of the elastic body set connected by the coincidence data point (elastic body shape data after deformation) is a solution of maximum likelihood as measurement data, and can be expected to be a value close to a true value.
That is, it is possible to realize a method of least squares in all degrees of freedom of data and output a result closer to the true value. And even if data with contradictions due to errors are connected by considering the measurement data as an elastic body, the contradictions due to the errors can be distributed to the entire measurement data as an elastic body by elastic deformation. Inconsistency in the connection region can be suppressed.

Now, a setting example will be described in which the deformation calculation unit 11d sets a plurality of elastic bodies (measurement data) so that the relative value becomes zero at the coincidence data point.
As shown in FIG. 3, the data value of the coincidence data point of the measurement data of the first measurement path (FIG. 3B) is (x _a , y _a , z _1a ), and the measurement data of the second measurement path ( In the case of the data value (x _a , y _a , z _2a ) of the matching data point in FIG. 3 (c)), the data value of each matching data point (each matching data element) is, for example, It is set to be an intermediate value (x _a , y _a , (z _1a + z _2a ) / 2) of data values.
In this case, the data value (x _a , y _a , z _1a ) of the coincidence data point (coincidence data element) of the measurement data of the first measurement path (FIG. 3B) is an intermediate value (x _a , y _a , The first measurement path measurement data regarded as an elastic body is elastically deformed so as to be displaced to the position of (z _1a + z _2a ) / 2), and coincides with the measurement data (FIG. 3C) of the second measurement path. The elastic body so that the data value (x _a , y _a , z _2a ) of the data point (matching data element) is displaced to the position of the intermediate value (x _a , y _a , (z _1a + z _2a ) / 2) The second measurement bus measurement data regarded as being elastically deformed.
As a result of the above processing, the data values of the matching data points (matching data elements) whose data values should match are matched (relative value is zero), and the matching data points are connected.
For the sake of simplicity of calculation, the matching data point data value should match _{(x a, y a, z} 1a), (x a, y a, z 2a) the data value x of the, y, of z , By changing only the values z _1a and z _{2a of the} stitching direction (here z direction) (by changing z _1a and z _2a to (z _1a + z _2a ) / 2), the data value may be matched. in this case, the value x _a which originally _coincide, y _a, since not changed, can be omitted operation of xy coordinates.
However, strictly speaking, if each elastic body is elastically deformed so that the values z _1a and z _2a in the stitching direction coincide with each other, the xy coordinate value of each coincidence data point of each elastic body also becomes the original value ( A value (x _{a ′} , y _{a ′} ) slightly deviated from x _a , y _a ). Therefore, when performing a strict calculation, the xyz coordinate values when the data values of the matching data points are matched should be (x _{a ′} , y _{a ′} , (z _1a + z _2a ) / 2). It is.
However, since the deviation in the xy coordinates is very small and hardly affects the calculation result, the calculation of the xy coordinate values may be omitted. That is, the xy coordinate value of the elastic body may use a value before connection of the coincidence data point, and may use a value after connection of the coincidence data point only for the z coordinate value.

Here, the displacement in the xy coordinates of each elastic body is obtained by sufficiently expanding each elastic body in the xy direction in advance and connecting the coincidence data points to elastically deform the elastic body. It can be suppressed by reducing the size to the original size. The expansion in the xy direction is made sufficiently larger than the displacement amount in the z direction of the displacement position of the elastic body accompanying the connection of the coincidence data points. By sufficiently expanding the elastic body in the xy direction, the displacement in the z direction caused by the connection of the coincidence data points becomes relatively small when viewed from the expanded elastic body, so that the displacement in the xy direction accompanying the connection is also small. become. By such a method, the influence of the compressive force and the tensile force acting on the elastic body can be suppressed, and more appropriate calculation becomes possible.

In the above description, the measurement data includes data elements (for example, (x _a , y _a , z _1a ) in FIG. 3B and (x _a , It was assumed that y _a , z _2a )) existed.
However, since the measurement data (data to be stitched) is discrete data, there is not always a data element that completely matches the measurement path intersection position of the actual measurement target. If high-precision data stitching is not required, data elements in the vicinity of the actual measurement path intersection position can be selected as matching data points from the data elements included in the measurement data. Is required, it is desirable to set the matching data points more accurately.

The coincidence data point setting unit 11c of the present embodiment also has a function of setting the coincidence data point more accurately when the measurement data does not include a data element that coincides with the actual measurement path intersection position.
FIG. 6 shows a more accurate way to set matching data points. Here, as shown in FIG. _6A , the first data element (x _1i , y _1i , z _1i ) and the second data element (x _{1 (i + 1)} , y _{1 ( i + 1), z 1 (} i + 1)) between the, and the third data element of the measurement data of the second measurement path _{(x 2j, y 2j, z} 2j) and the fourth data element _{(x 2 (j + 1)} , y 2 _Assume that an actual measurement path intersection (coincidence data point) exists between _{(j + 1)} and z2 _{(j + 1)} ).

In this case, the coincidence data point setting unit 11c uses the first data element (x _1i , y _1i , z _1i ) of the measurement data of the first measurement path and the first data based on the measurement data (discrete data) of the first measurement path. An interpolation function for interpolating a section between two data elements (x _{1 (i + 1)} , y _{1 (i + 1)} , z _{1 (i + 1)} ) (in this case, a primary interpolation function is adopted) and a second measurement path _Is interpolated between the third data element (x _2j , y _2j , z _2j ) and the fourth data element (x _{2 (j + 1)} , y _{2 (j + 1)} , z _{2 (j + 1)} ) of the measured data of An interpolation function (here, the primary interpolation function is minimized) is obtained.
Then, the coincidence data point setting unit 11c determines the position (x _a , y _a ) of the xy coordinates where the two interpolation functions intersect in the xy plane of FIG. 6A as the intersection in the xy coordinates.

Furthermore, the coincidence data point setting unit 11c obtains the value of z (= z _1a ) at the intersection xy coordinates (x _a , y _a ) using the interpolation function of the first measurement path. Thereby, the xyz coordinate value = (x _a , y _a , z _1a ) of the interpolation data element corresponding to the intersection point is obtained on the interpolation function of the first measurement path. The obtained interpolation data element (x _a , y _a , z _1a ) is added to the measurement data of the first measurement path and set as a coincidence data point in the measurement data of the first measurement path.
Similarly, the coincidence data point setting unit 11c obtains the value of z (= z _2a ) at the intersection xy coordinates (x _a , y _a ) using the interpolation function of the second measurement path. Thereby, the xyz coordinate value = (x _a , y _a , z _2a ) of the interpolation data element corresponding to the intersection is obtained on the interpolation function of the second measurement path. The obtained interpolation data element (x _a , y _a , z _2a ) is added to the measurement data of the second measurement path and set as a coincidence data point in the measurement data of the second measurement path.
With the above processing, more accurate matching data points can be set.

In this way, by setting the interpolated data elements interpolated between the data elements that make up the measurement data (data to be stitched) as matching data points, the matching data points are more accurate regardless of the sampling frequency of the measurement data. Can be set.

And when the deformation | transformation calculating part 11d makes the relative value of the set coincidence data point zero, as shown in FIG.6 (b), the data value of the interpolation data element set as a coincidence data point is each The first measurement path measurement data and the second measurement path measurement data regarded as an elastic body are elastically deformed so as to be displaced to the position of the intermediate value (x _a , y _a , z _3a ) of the data values of the coincidence data points. .

Returning to FIG. 5A, when the deformation calculation process by the finite element method (step S3) ends, the deformation calculation unit 11d obtains elastic body shape data of the elastic body aggregate after elastic deformation as stitching data. As a result of the deformation analysis by the finite element method, the deformation calculation unit 11d outputs not only the elastic body shape data of the elastic body aggregate after deformation but also data indicating the internal stress of the elastic body aggregate after deformation.
Based on the internal stress data of the deformed elastic body, the detection unit 11e of the processing unit 11 is improper measurement data (stitched data) such as a relatively large error or improperly set matching data. Detection processing for detecting a point (inappropriate connection of the elastic body) is executed (step S4).

For example, if the error included in one measurement data out of multiple measurement data is much larger than the error included in other measurement data for some reason, the measurement data with a large error It is a big contradiction to the data. If measurement data with a large error is connected to other measurement data to cause elastic deformation, the elastic measurement data with a large error will cause unreasonable elastic deformation. The stress becomes very large. That is, when there is a large discrepancy between a plurality of measurement data, the internal stress increases at a location where such measurement data is connected.

Therefore, the detection unit 11e uses the internal stress data to detect measurement data or coincident data points (elastic body connection positions) belonging to a range where the internal stress is large. As a result, more appropriate deformation analysis results can be obtained by excluding inappropriate measurement data or removing matching data points (connection of elastic bodies) and performing deformation analysis again. .

In order for the detection unit 11e to detect a range where the internal stress is large, for example, the internal stress is compared with a predetermined threshold, and a range where the internal stress exceeds the threshold can be detected as a range where the internal stress is large. The threshold value can be set, for example, based on the average value of the internal stress of the entire elastic body assembly (for example, the threshold value is set to twice the average value of the internal stress).

The output processing unit 11f of the processing unit 11 outputs the deformed elastic body shape data to the output unit (for example, a display unit such as a display) 14 as stitching data obtained by stitching a plurality of pieces of stitched data. The output unit 14 may output stitching data to the storage unit 12. FIG. 7 shows the output stitching data.

The RMS (Root Mean Square) of the measurement data before processing by the processing unit 11 is 54 nm and the PV (Peak-to-valley) is 510 nm (see FIG. 4), whereas the stitch according to Embodiment 11 is used. RMS in the stitching data after the bending process (elastic body shape data after deformation) was 15.8 nm, and PV was 100 nm. Therefore, it can be seen that the measurement accuracy of the stitching data is significantly improved as compared with the measurement data before the processing by the processing unit 11, and a highly accurate result is obtained.
Therefore, even when the measurement accuracy of the measurement device 2 is low, a highly accurate measurement result can be obtained by stitching.

Furthermore, in this embodiment, since the finite element method generally used for deformation analysis is used in the deformation calculation processing by the deformation calculation unit 11d, it is easy to construct a stitching device. Further, since the finite element method performs deformation analysis at high speed, stitching processing can be performed at high speed. For example, when a home computer is used as the processing device 3, the deformation calculation process (step S3) in the stitching process can be performed in a few seconds.

In addition, the measurement results (stitching data) shown in FIG. 7 are obtained when a circular disk having a smooth surface is measured, but in FIGS. 8 and 9, there are convex portions on the surface. The result when measuring the circular disk to be shown is shown.
FIG. 8 shows measurement data before processing by the processing unit 3 as in FIG. In the measurement data before processing, it is difficult to grasp that the convex portion exists at the position of the “convex portion that should exist” shown by the thick black line in FIG.
On the other hand, in the stitching data after processing by the processing unit 3, as shown in FIG. 9, a shape corresponding to the convex portion appears, and it can be seen that a highly accurate measurement result is obtained.
Also from FIG. 9, it can be confirmed that a highly accurate measurement result is obtained by the stitching process of the present embodiment.

Further, the elastic modulus setting unit 11b can obtain a measurement result with higher accuracy by changing the elastic modulus set for each of the plurality of measurement data according to the accuracy of each measurement data. For example, if the accuracy of the measurement data is high, the elastic modulus setting unit 11b sets a relatively large elastic modulus in the measurement data to make it difficult to deform, and if the accuracy of the measurement data is low, the elastic modulus setting unit 11b has a relatively small elastic modulus. Is set to the measurement data to facilitate deformation. In other words, highly reliable (high accuracy) measurement data is difficult to deform, while low reliability (low accuracy) measurement data is easily deformed, so that highly reliable measurement data is highly reliable. Since the deformation is performed along the measurement data, the accuracy of the stitching data can be increased.

For example, among the measurement paths in FIG. 2, the circular path is measured n times, and measurement data with small error and high accuracy (reliability) is obtained. On the other hand, for many linear paths, It is assumed that only one measurement is performed and only relatively low-precision measurement data is obtained. In this case, the elastic modulus set in the highly accurate circular path measurement data is made n times larger than the elastic modulus set in the low precision linear path measurement data, and the circular path measurement data (elasticity Make body shape data) difficult to deform.
Thereby, since the measurement data of the straight path with low accuracy is deformed in accordance with the measurement data of the circular path with high accuracy, high-precision stitching data is obtained. In addition, since the number of measurements can be reduced for the straight path, the measurement speed can be increased.

As for the accuracy (reliability) of the measurement data, the elastic modulus setting unit 11b may accept the number of times of measurement of each measurement data from the user via the input unit 13 as data relating to accuracy, or measurement. Data such as the number of measurements of individual measurement data may be acquired from the device 2.

FIG. 10 shows the evaluation process by the evaluation unit 11g. The evaluation process evaluates observation methods such as measurement and photographing. The evaluation unit 11g of the present embodiment evaluates the appropriateness of how to determine an observation range such as a measurement path (measurement range) and an imaging range.
The process of evaluating the appropriateness of how to determine the observation range is performed by eigenvalue analysis on a set of a plurality of elastic bodies connected to each other (step S11). The higher the eigenvalue (e.g., natural frequency) obtained as a result of the eigenvalue analysis, the higher the evaluation of how to determine the observation range is output (step S12).

A set of a plurality of elastic bodies connected to each other has a high rigidity as a whole when the eigenvalue is high, and is difficult to be deformed. If the eigenvalue is low, the rigidity is low as a whole and is easily deformed. The harder the set of elastic bodies, the more efficiently the data are linked together, and it can be evaluated that the data stitch is an excellent data stitch with a higher correction force. Therefore, the evaluation of the quality of the observation range of the data and the overlapping region can be performed in advance before observation by eigenvalue analysis.

FIG. 11 and FIG. 12 show that the eigenvalues of an elastic body set in which a plurality of linear elastic bodies corresponding to a plurality of measurement data are connected at the intersections of measurement paths differ depending on how to determine a linear measurement path (observation range). Show. The measurement path shown in FIG. 11A is the same as the measurement path shown in FIG. 2, and has a large number of straight paths and an annular path on the outer peripheral side of the straight paths. FIG. 11B is obtained by removing the circular path from the measurement path of FIG. Since the elastic body set corresponding to the measurement path in FIG. 11B does not have an elastic body corresponding to the annular path, the eigenvalue is low and the rigidity is low.

FIGS. 12A to 12C also show the measurement paths. The eigenvalues become smaller in the order of FIGS. 12 (a), (b), and (c). Therefore, the data correction force of the measurement path in FIG. 12A is the highest, and the data correction force decreases in the order of FIGS. 12B and 12C.

FIG. 13 shows how to determine the shooting range (observation range) of image data, and that the eigenvalues of the elastic body sets in which a plurality of planar elastic bodies corresponding to a plurality of image data are connected in overlapping areas of the shooting range are different. Yes. FIGS. 13A and 13B show a state in which 16 pieces of image data are overlapped in the overlapping region of the shooting range. In FIGS. 13A and 13B, the overlapping region is indicated by hatching. The overlapping area of the imaging range shown in FIG. 13B is larger than the overlapping area in FIG. Therefore, the eigenvalue is higher in FIG. 13B, and the image data obtained according to the imaging range set as shown in FIG. 13B has a higher data correction capability by data stitching.

However, since the overlapping area is large in FIG. 13B, the range that can be covered by 16 pieces of image data is small in FIG. 13B. In FIG. 13B, the range that can be covered by the 16 images in FIG. 13A is indicated by “A”. In FIG. 13B, if an attempt is made to secure a range having the same size as the range A, the required number of image data increases and the amount of data increases. An increase in data amount leads to an increase in data acquisition cost.
Therefore, when evaluating how to determine the imaging range (observation range), an evaluation value that considers not only the data correction power based on the eigenvalue but also the necessary data amount may be obtained.
Further, since the density of data elements as viewed from the whole measurement target (observation target) varies depending on how the measurement path (observation range) is set, the density of the data elements may be considered in the evaluation value. For example, in FIGS. 12A and 12B, while FIG. 12A has a large number of measurement paths and a high data density, the data density decreases in the order of FIGS. 12B and 12C. The part which is not measured in a measurement object will increase. Therefore, the higher the data duplication density, the more accurate the data can be evaluated.

In the evaluation process, the data regarded as the elastic body may be actual observation data or evaluation data created in accordance with the observation range. For example, the evaluation data for the measurement data having the data structure shown in FIGS. 3B and 3C can be such that the height (z coordinate value) is all set to a constant value. Further, the evaluation data for the image data can be obtained by setting all the pixel values to a constant value.

Hereinafter, the data stitching according to the embodiment will be described in more detail with an explanation from a mathematical viewpoint. FIG. 14 shows a plurality of data (first data and second data) to be stitched. Here, for simplification of explanation, each data element _is assumed to have the _{x m_n} and ^f n _{(x m_n)} of 2-dimensional value (discrete value). 14A, the x coordinate value of the data shown in FIG. 3 is represented by _{xm_n} , the y coordinate value of the data shown in FIG. 3 is omitted, and the z coordinate value of the data shown in FIG. This corresponds to that represented by f ⁿ (x _{m —} ⁿ ).
In FIG. 14 (a),
n = 1 to N
N: Number of data to be stitched (data to be stitched) m_n = 1 to M_n
M_n: the number of data elements constituting the nth data.

The x value of each data element of the first data with n = 1 is denoted by x _{m — 1} , and the f value of each data element of the first data is denoted by f ¹ (x _{m — 1} ). The x value of each data element of the second data for which n = 2 is indicated by x _{m — 2} , and the f value of each data element of the second data is indicated by f ² (x _{m — 2} ).

As shown in FIG. 14A, the first data and the second data are discrete data. When the discrete first data and second data are stitched, each piece of data is regarded as an elastic body having a continuous shape. The shape of each elastic body is a continuous shape connecting values (x _{m —} ⁿ , f ⁿ (x _{m —} ⁿ )) indicating data elements constituting each data.
FIG. 14B shows a line shape in which data elements are connected by lines for each data. The shape of the elastic body corresponding to the discrete first data is represented by the function f ¹ (x). The line shape of the elastic body corresponding to the discrete second data is represented by the function f ² (x). Hereinafter, the function f ⁿ (x) indicates the shape of the elastic body when the n-th data is regarded as the shape of the elastic body.

Here, FIG. 14B shows a curved shape in which each data element is connected by a smoothly bent line. On the other hand, FIG. 14C shows a polygonal line shape in which each data element is connected by a straight line. When the stitched data is regarded as elastic body shape data, as the shape of the elastic body, FIG. 14B may be considered, or FIG. 14C may be considered. As will be described later, in the data stitching using elastic deformation, it is sufficient to handle the displacement u ⁿ (x) of each data element. Therefore, even if the shape of the elastic body is FIG. FIG. 14C is also equivalent. When the elastic body deformation model is a finite element model composed of beam elements, the data elements are connected by linear beam elements, so the shape of the elastic body deformation model is as shown in FIG. .
However, here, for convenience of explanation from a mathematical point of view, the shape of the elastic body is a shape in which data elements are connected by a smooth curve as shown in FIG.

For generalization, FIG. 15 shows the shape function corresponding to the first data as f ⁱ (x) (i = 1), and the shape function corresponding to the second data as f ^j (x) ( j = 2). Note that i ≠ j.
Here, the matching data point of the first data is set to f ⁱ (x _{2 —} ⁱ ), and the matching data point of the second data is set to f ^j (x _{2 —} ^j ). The connection of two elastic bodies at the displacement position corresponding to the coincidence data point is mathematically expressed as f ⁱ (x _{2 —} ⁱ ) such that f ⁱ ′ (x _{2 —} ⁱ ) = f ^j ′ (x _{2 —} ^j ). f ^j (x _{2 —} ^j ) is displaced. In addition, f ⁱ ′ (x _{2 —} ⁱ ) and f ^j ′ (x _{2 —} ^j ) indicate values after displacement of f ⁱ (x _{2 —} ⁱ ) and f ^j (x _{2 —} ^j ), respectively.
That is, the generalized stitching condition is as the following formula (1).

f ⁱ 'represents the amount of displacement of the _{(x 2_i)} ^u i and _{^{(x 2_i), f j (}} x 2_j) from ^{f j'} _(x 2_i) from ^{f i} the displacement amount of the _{(x 2_j)} ^u j ₍ x 2 — _j ), the stitching condition shown in Expression (1) is expressed as follows.

The stitching conditions shown in Equation (1) and Equation (2) correspond to connecting a plurality of elastic bodies to each other at the displacement position (matching data point).

If _f ⁱ _(x 2_i) and ^f j a _{(x 2_j)} regarded as an elastic _{^{body, f i '(x 2_i)}} = f j' such that the ^{_{(x 2_j), f i (}} x 2_i) and ^f j ( When x 2 — _j ) is displaced, the shape indicated by the functions f ¹ (x) and f ² (x) changes as a whole.
FIG. 16 shows functions f ⁱ ′ (x) and f ^j ′ (x) corresponding to the shape of the elastic body after deformation. The function f ⁱ ′ (x) indicates a function f ⁱ (x) deformed with the connection of two elastic bodies. The function f ^j ′ (x) indicates a function f ^j (x) that is deformed along with the connection of two elastic bodies.
That is, it is as the following formulas (3) and (4).

As shown in FIG. 17, u ⁱ (x) is a function (displacement function) indicating the amount of displacement from f ⁱ (x) to f ⁱ ′ (x), and u ^j (x) is f ^j (x It is a function (displacement function) indicating the amount of displacement from x) to f ^j ′ (x).
Here, there are innumerable ways to modify the functions f ⁱ (x) and f ^j (x) that satisfy the stitching condition. Accordingly, there are innumerable displacement functions u ⁱ (x) and u ^j (x). Therefore, it is necessary to optimize the displacement functions u ⁱ (x) and u ^j (x).
In order to optimize the displacement function, in the present embodiment, the transformation from the function f ⁱ (x), f ^j (x) to the function f ⁱ ′ (x), f ^j ′ (x) is the function f ⁱ (x ), F ^j (x). An elastic body assembly made up of a plurality of elastic bodies connected to each other takes a shape that minimizes elastic energy caused by bending. By considering that the deformation of the function is elastic deformation, the displacement functions u ⁱ (x) and u ^j (x) can be optimized by a method of analyzing the elastic deformation.
Minimizing the elastic energy by bending is equivalent to minimizing the integral value of the curvature of the displacement functions u ⁱ (x), u ^j (x) in mathematical terms. Therefore, the optimization of the displacement function when the concept of elastic deformation is introduced is formulated, for example, as in the following equation (5).

In equation (5), the sum of integral values of the squares of the second-order differential values (curvatures) of the displacement functions u ⁿ (x) (n = 1 to N) is minimized. In the present embodiment, the process of “obtaining elastic deformation” can be performed by optimizing the displacement function according to the concept according to the equation (5). The finite element method described above is also one of the processes based on the concept according to Equation (5). The finite element method is excellent as a method for obtaining an elastic body shape that minimizes elastic energy caused by bending. Therefore, the finite element method is very suitable as a method for obtaining elastic deformation.

　式（６）では、変位関数ｕ^ｎ（ｘ）（ｎ＝１～Ｎ）それぞれの二階微分値（曲率）に重みｗ_ｎを乗じている。データの精度（信頼度）が高いほど、弾性率、すなわち重みｗ_ｎを大きくする。これにより、高い精度を持つデータに対応した弾性体に対応する変位関数の曲率がより大きく評価される（弾性体が変形しにくい）ことになる。 Optimization in the case where different elastic moduli are set for a plurality of elastic bodies is formulated as shown in Equation (6).

In equation (6), it is multiplied by the weight _{w n} to the displacement function ^{u n (x) (n =} 1 ~ N) each second-order differential value (curvature). Higher data accuracy (reliability) is high, the elastic modulus, i.e. the weight w _n increases. As a result, the curvature of the displacement function corresponding to the elastic body corresponding to the data with high accuracy is more greatly evaluated (the elastic body is not easily deformed).

When the optimized (elastically deformed) displacement function u ⁿ (x) (n = 1 to N) is obtained, the original data f ⁿ (x _{m —} ⁿ ) and the optimized displacement function are calculated based on the equation (7). f ^{n ′} (x _{m —} ⁿ ) is obtained from u ⁿ (x). As shown in FIG. 18, f ^{n ′} (x) indicates the shape of the elastic body that is elastically deformed.

The present invention is not limited to the above embodiment. For example, the stitching data to be stitched is not limited to measurement data, and may be arbitrary data such as image data (satellite photo data, panoramic photo data, etc.). When the image is a color image, data stitching may be performed for each of the RGB colors.
Further, since the finite element method can be applied to data of two dimensions or less, naturally, the stitched data may be data of two dimensions or less.
Further, the stitched data may be four-dimensional data or more. If the stitched data is data of four or more dimensions, the stitched data is regarded as elastic body shape data of four or more dimensions, and the finite element method expanded to a dimension corresponding to the number of data dimensions is used. Good.
That is, the stitching data may be arbitrary n-dimensional data (n is an integer of 1 or more).

Further, the elastic modulus set in the stitched data may be different for each dimension. For example, in the case of four-dimensional data composed of three-dimensional coordinates and time, since the properties of the coordinates and time are different, the elastic modulus related to coordinates and the elastic modulus related to time are made different so that the deformation of coordinates and the deformation of time are different. May be.

When the stitched data is image data, the coincidence data point may be set by user input, or a common object in a plurality of images is extracted by image processing, and the extracted portions are matched. It may be set as a data point.

The deformation analysis method is not limited to the finite element method, and may be another method capable of obtaining the deformation by numerical analysis.

DESCRIPTION OF SYMBOLS 1 Stitching apparatus 2 Data acquisition part 3 Processing apparatus 11 Processing part 11a Data acquisition processing part 11b Elastic modulus setting part 11c Matching data point setting part 11d Deformation calculation part 11e Detection part 11f Output processing part 11g Evaluation part 12 Storage part 13 Input part 14 Output section

Claims (19)

A data stitching device for stitching a plurality of data, each having a plurality of data elements,
Each of the plurality of data is regarded as data of the shape of the elastic body, and includes a processing unit that obtains elastic deformation generated in each of the plurality of elastic bodies corresponding to the plurality of data,
The shape of the elastic body is a shape connecting values indicated by a plurality of data elements included in the data,
The elastic deformation is caused by connecting the plurality of elastic bodies to each other so that the positions coincide with each other at the position of the elastic body corresponding to the data element that should have the same data element value in the plurality of data. And
The data processing apparatus, wherein the processing unit outputs data indicating the shapes of the plurality of elastic bodies having undergone the elastic deformation as stitching data obtained by stitching the plurality of data. The data stitching device according to claim 1, wherein the processing unit is configured to set an elastic modulus of each of the plurality of elastic bodies. The data stitching apparatus according to claim 2, wherein the processing unit can set different elastic moduli for each of the plurality of elastic bodies. The data stitching device according to claim 2 or 3, wherein the processing unit is configured to determine an elastic modulus of each of the plurality of elastic bodies based on reliability of each of the plurality of data. The processing unit is configured to perform a process of adding an interpolation data element that interpolates the plurality of data elements to at least one of the plurality of data.
The data stitching device according to any one of claims 1 to 4, wherein a data element whose data element values should match in the plurality of data includes the interpolated data element. 2. The processing unit is configured to obtain internal stress of the plurality of elastic bodies in which the elastic deformation has occurred, and to detect inappropriate data among the plurality of data based on the internal stress. 6. The data stitching device according to any one of 1 to 5. The processing unit obtains internal stresses of the plurality of elastic bodies in which the elastic deformation has occurred, and detects inappropriate connection of the plurality of elastic bodies based on the internal stresses. 2. A data stitching apparatus according to item 1. The data stitching device according to any one of claims 1 to 7, wherein the processing unit obtains the elastic deformation by deformation analysis using a plurality of elastic body deformation models generated based on the plurality of data. . The data stitching apparatus according to claim 8, wherein each of the plurality of elastic body deformation models includes a plurality of elements that can be bent and deformed at least. The data stitching device according to claim 9, wherein the element includes a beam element. The data stitching device according to claim 9, wherein the element includes a shell element. The data stitching apparatus according to any one of claims 8 to 12, wherein the deformation analysis is performed by a finite element method. The data stitching device according to any one of claims 1 to 12, wherein the elastic deformation includes at least bending deformation. The plurality of data is a plurality of shape measurement data obtained by measuring the surface of the measurement object along a plurality of measurement paths,
Each of the plurality of data elements is a data element indicating a measurement value on the measurement path,
Each of the plurality of measurement paths has an intersection with another measurement path included in the plurality of measurement paths,
The shape of the elastic body is a shape of a line connecting the measurement values indicated by a plurality of data elements included in the shape measurement data,
The elastic deformation is an elastic deformation generated by connecting the plurality of elastic bodies so that the positions coincide with each other at a position of the elastic body corresponding to a data element indicating a measurement value of the intersection. 14. The data stitching device according to any one of items 13. The plurality of data is image data,
Each of the plurality of data elements is a pixel;
Each of the shooting ranges of the plurality of image data has an overlapping area with shooting ranges of other image data included in the plurality of image data,
The shape of the elastic body is a shape of a surface connecting pixel values indicated by a plurality of pixels included in the image data,
The elastic deformation is an elastic deformation generated by connecting the plurality of elastic bodies so that the positions coincide with each other at a position of the elastic body corresponding to a pixel included in the overlapping region. The data stitching apparatus of any one of Claims. The plurality of data is a plurality of observation data obtained by observing an observation object with an observation device,
The data stitching apparatus according to any one of claims 1 to 13, wherein each of the observation ranges of the plurality of observation data has an overlapping area with an observation range of other observation data included in the plurality of observation data. The processing unit is configured to perform a process of evaluating how to observe the observation object,
The data stitching apparatus according to any one of claims 1 to 16, wherein the evaluation process is performed based on a result of eigenvalue analysis on the entire plurality of elastic bodies. A method in which a computer performs a process of stitching a plurality of data each having a plurality of data elements,
Each of the plurality of data is regarded as data of the shape of the elastic body, and the computer executes a process for obtaining elastic deformation occurring in each of the plurality of elastic bodies corresponding to the plurality of data,
The computer outputs data indicating the shapes of the plurality of elastic bodies in which the elastic deformation has occurred as stitching data obtained by stitching the plurality of data,
The shape of the elastic body includes a value connecting values indicated by a plurality of data elements included in the data,
The elastic deformation is caused by connecting the plurality of elastic bodies to each other so that the positions coincide with each other at the position of the elastic body corresponding to the data element that should have the same data element value in the plurality of data. Is the method. A computer program for causing a computer to execute a process of stitching a plurality of data each having a plurality of data elements,
The processing is as follows:
Each of the plurality of data is regarded as data of the shape of the elastic body, obtaining elastic deformation generated in each of the plurality of elastic bodies corresponding to the plurality of data;
Outputting data indicating the shapes of the plurality of elastic bodies in which the elastic deformation has occurred, as stitching data obtained by stitching the plurality of data;
Including
The shape of the elastic body is a shape connecting values indicated by a plurality of data elements included in the data,
The elastic deformation is caused by connecting the plurality of elastic bodies to each other so that the positions coincide with each other at the position of the elastic body corresponding to the data element that should have the same data element value in the plurality of data. Is a computer program.

Images