Information processing method, information processing device, and program

Abstract

This information processing device acquires an image, generates a data point group corresponding to an object included in the image, calculates an approximate line by fitting related to the data point group, divides an entire region of the image into a plurality of partial regions in calculation of the approximate line, and calculates a partial approximate line, which is a part of the approximate line, by fitting targeting a plurality of data points included in the partial region in the data point group.

Description

Information processing method, information processing device, and program

[0001] The present disclosure relates to an information processing method, an information processing device, and a program.

[0002] Patent Document 1 discloses a time series data analysis device according to the background art, which divides time series data into a plurality of time domains and obtains a regression line for each time domain.

[0003] According to the background art, there has been no consideration given to improving the accuracy of curve fitting for a data point group generated by instance segmentation of an image.

[0004] JP 2013-89112 A

[0005] An object of the present disclosure is to provide an information processing method, an information processing device, and a program that can improve the accuracy of fitting related to a data point group.

[0006] In an information processing method according to one aspect of the present disclosure, an information processing device acquires an image, generates a data point cloud corresponding to an object included in the image, calculates an approximation line by fitting the data point cloud, and in calculating the approximation line, divides the entire area of ​​the image into a plurality of partial areas, and calculates a partial approximation line that is a part of the approximation line by fitting a plurality of data points included in the partial areas of the data point cloud.

[0007] FIG. 1 is a simplified diagram showing the configuration of a computer device according to an embodiment of the present disclosure. FIG. 2 is a simplified diagram showing functions of a processing unit. FIG. 3 is a flowchart showing processing details executed by the processing unit. FIG. 4 is a flowchart showing processing details executed by the processing unit. FIG. 5 is a simplified diagram showing an example of an acquired image. FIG. 6 is a simplified diagram showing an example of an image including a data point cloud. FIG. 7 is a simplified diagram showing an example of calculation of an overall approximation line. FIG. 8 is a simplified diagram showing a first example of an image divided into regions. FIG. 9 is a simplified diagram showing a second example of an image divided into regions. FIG. 10 is a simplified diagram showing a third example of an image divided into regions. FIG. 11 is a simplified diagram showing an example of calculation of a partial approximation line. FIG. 12 is a diagram showing how partial regions are superimposed on each other. FIG. 13 is a flowchart showing processing details executed by the processing unit. FIG. 14 is a schematic diagram showing a situation in which region division progresses in the reduction direction.

[0008] (Knowledge that forms the basis of this disclosure) In object detection, which detects curved or straight line objects contained in an image, instance segmentation of the image is performed to generate a data point group indicating pixels corresponding to the object, and an approximation line such as a regression curve is calculated by curve fitting the data point group of the entire area of ​​the image.

[0009] However, depending on the shape of the object, for example, if the object includes a first part and a second part that have the same X coordinate but significantly different Y coordinates, the accuracy of curve fitting targeting the data point group of the entire image area may decrease.

[0010] In order to solve this problem, the inventor discovered that the accuracy of fitting can be improved by dividing the entire area of ​​an image into multiple partial areas and performing fitting for each partial area, and came up with the present disclosure.

[0011] Next, each aspect of the present disclosure will be described.

[0012] In an information processing method according to a first aspect of the present disclosure, an information processing device acquires an image, generates a data point cloud corresponding to an object included in the image, calculates an approximation line by fitting the data point cloud, and in calculating the approximation line, divides the entire area of ​​the image into a plurality of partial areas, and calculates a partial approximation line that is a part of the approximation line by fitting a plurality of data points included in the partial areas of the data point cloud.

[0013] According to the first aspect, the accuracy of fitting for the data point group can be improved.

[0014] In the information processing method according to the second aspect of the present disclosure, in the first aspect, the fitting may include nonlinear regression analysis, and the approximation line may include a regression curve.

[0015] According to the second aspect, the accuracy of calculation can be improved in calculating a regression curve by nonlinear regression analysis.

[0016] In the information processing method according to the third aspect of the present disclosure, in the first or second aspect, in generating the data point group, the data point group corresponding to the shape of the object may be generated by instance segmentation of the image.

[0017] According to the third aspect, a data point group corresponding to the shape of an object can be appropriately generated by instance segmentation of an image.

[0018] In the information processing method according to a fourth aspect of the present disclosure, in any one of the first to third aspects, the object may include a curve or a straight line included in the image.

[0019] According to the fourth aspect, it is possible to realize object detection targeting curved or straight line objects included in an image.

[0020] In the information processing method according to the fifth aspect of the present disclosure, in any one of the first to fourth aspects, in calculating the approximation line, the accuracy of fitting for the data point group is calculated, and if the accuracy is less than an allowable value, the region division and calculation of the partial approximation line are performed.

[0021] According to the fifth aspect, when the accuracy is equal to or greater than the allowable value, it is possible to avoid unnecessary region division and calculation of partial approximation lines.

[0022] In the information processing method according to a sixth aspect of the present disclosure, in any one of the first to fifth aspects, the region division may include dividing the entire region of the image into the plurality of partial regions of a predetermined size.

[0023] According to the sixth aspect, the area division process can be easily executed.

[0024] In an information processing method according to a seventh aspect of the present disclosure, in a sixth aspect, when the image has orthogonal first and second coordinates, and the data point group includes a first data point and a second data point having a common coordinate value of the first coordinate and a coordinate value of the second coordinate that differs by more than a threshold value, the specified size may be set to a size that includes the first data point and the second data point in different partial areas.

[0025] According to the seventh aspect, the accuracy of fitting for the data point group can be further improved.

[0026] An information processing method according to an eighth aspect of the present disclosure is the sixth or seventh aspect, wherein the shape of the partial region is preferably set based on an aspect ratio of the image or a distribution shape of the data point cloud.

[0027] According to the eighth aspect, the accuracy of fitting for the data point group can be further improved.

[0028] In an information processing method according to a ninth aspect of the present disclosure, in any one of the first to eighth aspects, in the region division, it is preferable that the entire region of the image is divided into the plurality of partial regions of optimal size by updating the size of the partial regions.

[0029] According to the ninth aspect, the accuracy of fitting for the data point group can be further improved.

[0030] An information processing method according to a tenth aspect of the present disclosure is, in the ninth aspect, preferably wherein the size update is repeated in the shrinking direction until the accuracy of fitting for the data point group becomes equal to or greater than an allowable value, and the optimal size is the size of the partial region at which the accuracy becomes equal to or greater than the allowable value.

[0031] According to the tenth aspect, the accuracy of fitting for the data point group can be improved to an acceptable value or more.

[0032] An information processing method according to an eleventh aspect of the present disclosure is the ninth or tenth aspect, wherein the shape of the partial region is set based on an aspect ratio of the image or a distribution shape of the data point cloud.

[0033] According to the eleventh aspect, the accuracy of fitting for the data point group can be further improved.

[0034] In an information processing method according to a twelfth aspect of the present disclosure, in any one of the first to eleventh aspects, in the region division, it is preferable that the entire region of the image is divided into the plurality of partial regions by overlapping parts of adjacent partial regions.

[0035] According to the twelfth aspect, the accuracy of fitting for the data point group can be further improved.

[0036] An information processing device according to a thirteenth aspect of the present disclosure includes an information processing circuit that acquires an image, generates a data point cloud corresponding to an object included in the image, calculates an approximation line by fitting the data point cloud, and in calculating the approximation line, divides the entire area of ​​the image into a plurality of partial areas, and calculates a partial approximation line that is a part of the approximation line by fitting a plurality of data points included in the partial areas of the data point cloud.

[0037] According to the thirteenth aspect, the accuracy of fitting for the data point group can be improved.

[0038] A program according to a fourteenth aspect of the present disclosure is a program for causing an information processing device to execute a process, the process comprising acquiring an image, generating a data point group corresponding to an object included in the image, calculating an approximation line by fitting the data point group, and in calculating the approximation line, dividing the entire area of ​​the image into a plurality of partial areas, and calculating a partial approximation line that is a part of the approximation line by fitting a plurality of data points included in the partial areas of the data point group.

[0039] According to the fourteenth aspect, the accuracy of fitting for the data point group can be improved.

[0040] The present disclosure can also be realized as a program that causes a computer to execute each characteristic configuration included in such a method or apparatus, or as a system operated by this program. Needless to say, such a computer program can be distributed on a computer-readable non-transitory recording medium such as a CD-ROM or via a communication network such as the Internet.

[0041] (Embodiments of the Present Disclosure) Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. Elements with the same reference numerals in different drawings indicate the same or corresponding elements. Furthermore, the components, the arrangement positions of the components, the connection forms, the order of operations, etc. shown in the following embodiments are merely examples and are not intended to limit the present disclosure. The present disclosure is limited only by the claims. Therefore, among the components in the following embodiments, components that are not described in the independent claims that represent the highest concept of the present disclosure are not necessarily required to achieve the objectives of the present disclosure, but are described as constituting more preferred forms.

[0042] 1 is a diagram showing a simplified configuration of a computer device 1 according to an embodiment of the present disclosure. The computer device 1 is a general-purpose personal computer. However, the computer device 1 may be a dedicated terminal or a server device such as an edge server or a cloud server.

[0043] The computer device 1 includes an input unit 11, a processing unit 12, a display unit 13, a storage unit 14, and a communication unit 15. The input unit 11 includes an input device such as a keyboard or a mouse. The processing unit 12 (information processing device or information processing circuit) includes a processor such as a CPU. The display unit 13 includes a display device such as a liquid crystal display or an organic EL display. The storage unit 14 includes a storage device such as a HDD, SSD, or semiconductor memory. The communication unit 15 includes a communication module compatible with a communication protocol such as IP.

[0044] The storage unit 14 stores the program 20. The storage unit 14 has a computer-readable non-volatile storage medium such as a ROM, and the program 20 is stored in the storage medium.

[0045] 2 is a simplified diagram showing the functions of the processing unit 12. The processing unit 12 has an acquisition unit 31, a generation unit 32, an analysis unit 33, and an output unit 34. The analysis unit 33 includes a division unit 41 and a calculation unit 42. The processing content by executing each function will be described later. These functions may be realized by a processor executing a program 20 read from the storage unit 14, or may be realized using a dedicated circuit such as an ASIC or FPGA.

[0046] 3 and 4 are flowcharts showing the processing contents executed by the processing unit 12. Fig. 4 shows a subroutine of step SP3 shown in Fig. 3.

[0047] First, in step SP1, the acquisition unit 31 acquires the processing target image 50. The acquisition unit 31 acquires the image 50 via the communication unit 15 from a camera that captures the image 50, for example.

[0048] FIG. 5 is a simplified diagram illustrating an example of an acquired image 50. The image 50 is, for example, an image captured by a wide-angle camera mounted on an autonomous vehicle traveling along an aisle 62 in a warehouse. A placement area 64 for an item 65 is defined at the end of the aisle 62 within a certain distance from a wall 61. The boundary of the placement area 64 is indicated by a straight white line 63 painted on the aisle 62. However, because the image 50 was captured by the wide-angle camera, the curved white line 63 is included in the image 50 due to distortion. In the example shown in FIG. 5, the item 65 is placed across a portion of the white line 63. As a result, the item 65 divides the white line 63 into two regions. The image 50 has an X-axis extending horizontally and a Y-axis extending vertically. The acquisition unit 31 stores the acquired image 50 in the storage unit 14.

[0049] 3 , next, in step SP2, the generation unit 32 generates a data point cloud 70 corresponding to the shape of an object included in the image 50 by instance segmentation of the image 50. This makes it possible to appropriately generate the data point cloud 70 corresponding to the shape of the object by instance segmentation of the image 50. In the example of this embodiment, the object to be detected is a white line 63. However, the object may be a curved or straight line other than the white line 63. This makes it possible to realize object detection targeting a curved or straight line object included in the image 50.

[0050] FIG. 6 is a simplified diagram illustrating an example of an image 50 including a data point cloud 70. The data point cloud 70 includes a plurality of data points, each of which corresponds to a pixel on or near a white line 63. For example, the image 50 acquired in step SP1 is input into an AI model such as a machine-learned CNN, and the image 50 including the data point cloud 70 is output from the AI ​​model. The data point cloud 70 is also divided into two regions corresponding to the shape of the white line 63, which is divided into two regions. Note that, because it is difficult to accurately recognize thin linear objects in instance segmentation, the image 50 including the data point cloud 70 is prone to missing or misaligned data points relative to the white line 63.

[0051] Referring to FIG. 3 , next, in step SP3, the analysis unit 33 calculates an approximation line by fitting the data point group 70. The fitting includes curve fitting using nonlinear regression analysis. The approximation line includes a regression curve. This can improve the calculation accuracy in calculating the regression curve using nonlinear regression analysis. Note that the fitting may include linear regression analysis, and the approximation line may include a regression line.

[0052] 4, in step SP31, the calculation unit 42 calculates an overall approximation line 71 by fitting a data point group 70 relating to the entire region of the image 50. The fitting may be curve fitting using a polynomial curve, or estimation of an optimal function using the least squares method.

[0053] 7 is a simplified diagram showing an example of calculation of an overall approximation line 71. A data point cloud 70 includes a first portion P1 and a second portion P2 that have a common X coordinate but significantly different Y coordinates. Because the overall approximation line 71 passes between the first portion P1 and the second portion P2, the fitting accuracy is particularly low in this portion.

[0054] 4, next, in step SP32, the calculation unit 42 calculates the fitting accuracy of the overall approximation line 71. The fitting accuracy of the overall approximation line 71 corresponds to the sum of the distances between the overall approximation line 71 and each data point, and for example, the sum of squared residuals can be used. The smaller the sum of squared residuals, the higher the accuracy, and the larger the sum of squared residuals, the lower the accuracy.

[0055] Next, in step SP33, the division unit 41 determines whether the accuracy calculated in step SP32 satisfies the tolerance condition (i.e., whether the accuracy is equal to or greater than a predetermined tolerance). The tolerance may be a preset fixed value or a variable value that varies depending on the image 50, etc.

[0056] If the accuracy satisfies the tolerance condition (i.e., if the accuracy is equal to or greater than the tolerance value; step SP33: YES), the process returns to the main routine. This makes it possible to avoid unnecessary steps SP34 to SP36 being executed when the accuracy satisfies the tolerance condition.

[0057] If the accuracy does not satisfy the tolerance condition (i.e., if the accuracy is less than the tolerance value; step SP33: NO), then in step SP34, the division unit 41 divides the entire area of ​​the image 50 into multiple partial areas 80 (i.e., spatially divides it).

[0058] 8 is a simplified diagram showing a first example of an image 50 after region division. In the example of FIG. 8, the division unit 41 divides the image 50 into six parts in the X direction and four parts in the Y direction, thereby dividing the entire region of the image 50 into 24 partial regions 80 of a predetermined size. This allows the region division process to be easily performed. The partial regions 80 are the partial region 80 in the first row and first column. １１ to partial area 80 in the fourth row and sixth column ４６ In the example of FIG. 8, each partial region 80 has a square shape.

[0059] The division is not limited to 24 divisions, i.e., 6 divisions in the X direction and 4 divisions in the Y direction, but may be any division. The division may be more coarse than 24 divisions or more fine than 24 divisions.

[0060] However, if the data point group 70 includes a first data point (first part P1) and a second data point (second part P2) that have a common X-coordinate value but whose Y-coordinate value differs by more than a threshold value, the size of the partial area 80 may be set so that the first data point and the second data point are included in different partial areas 80.

[0061] 9 is a simplified diagram showing a second example of the region-divided image 50. In the example of FIG. 9, the division unit 41 divides the image 50 into three in the X direction and two in the Y direction, thereby dividing the entire region of the image 50 into six partial regions 80. In this case, the partial region 80 in which the first data point with coordinate values ​​(X1, Y1) and the second data point with coordinate values ​​(X1, Y2) are the same is selected. １１, and the size of the partial region 80 can be said to be too large. Note that the difference between the coordinate value Y1 and the coordinate value Y2 is greater than the threshold value. Therefore, in this case, it is desirable for the dividing unit 41 to set the size of the partial region 80 as shown in the example of FIG. 8. This can further improve the accuracy of fitting for the data point cloud 70.

[0062] 8 and 9. The dividing unit 41 may set the shape of the partial region 80 based on the aspect ratio of the image 50 or the distribution shape of the data point cloud 70.

[0063] Fig. 10 is a simplified diagram showing a third example of the region-divided image 50. In the example of Fig. 10, the division unit 41 sets the shape of the partial regions 80 to a horizontally long rectangle in accordance with the horizontal aspect ratio of the image 50. In addition, in this example, the division unit 41 divides the image 50 into three in the X direction and three in the Y direction, thereby dividing the entire region of the image 50 into nine partial regions 80. This can further improve the accuracy of fitting for the data point cloud 70.

[0064] 4 , next, in step SP35, the calculation unit 42 calculates a partial approximation line corresponding to each partial region 80 by fitting a plurality of data points included in each partial region 80 of the data point group 70. The fitting may be curve fitting using a polynomial curve, or estimation of an optimal function by the least squares method.

[0065] 11 is a simplified diagram showing an example of calculation of a partial approximation line. １１ By curve fitting to the plurality of data points included in １１ Partial approximation line 71 corresponding to １１ Furthermore, the calculation unit 42 calculates the partial region 80 １２ By curve fitting to the plurality of data points included in １２ Partial approximation line 71 corresponding to １２Similarly, for other partial regions 80 containing multiple data points, the calculation unit 42 calculates a partial approximation line corresponding to each partial region 80 by curve fitting the multiple data points contained in each partial region 80.

[0066] Referring to FIG. 4, next in step SP36, the dividing unit 41 generates an approximation line spanning a plurality of partial regions 80 by joining consecutive partial approximation lines together.

[0067] Referring to FIG. 11, the dividing unit 41 divides the partial approximation line 71 １１ and partial approximation line 71 １２ By combining the above, a plurality of partial regions 80 １１ , 80 １２ Similarly, the dividing unit 41 generates an approximation line 71A that spans a plurality of partial regions 80.

[0068] 3, next, in step SP4, the output unit 34 outputs result data indicating the approximation lines 71A and 71B generated in step SP3. The processing unit 12 may display a predetermined alert or the like generated based on the result data on the display unit 13.

[0069] According to this embodiment, in calculating the approximation line 71, the processing unit 12 divides the entire area of ​​the image 50 into a plurality of partial areas 80, and calculates the approximation line 71 by dividing the entire area of ​​the image 50 into a plurality of partial areas 80. １１ , 80 １２ A partial approximation line 71A is obtained by fitting a plurality of data points included in the approximation line 71B. １１ , 71 １２ This makes it possible to improve the accuracy of fitting for the data point group 70.

[0070] Various modifications of the embodiments of the present disclosure will be described below. The modifications described below can be applied in any combination.

[0071] (First Modification) The dividing unit 41 may divide the entire area of ​​the image 50 into a plurality of partial areas 80 by overlapping parts of adjacent partial areas 80 with each other.

[0072] FIG. 12 is a diagram showing how partial regions 80 overlap each other.

[0073] The dividing unit 41 sets an intermediate partial region 80 for each partial region 80 at a position shifted by half the size of the partial region 80. Specifically, the dividing unit 41 sets an intermediate partial region 80UL between the central partial region 80 and the partial region 80 adjacent to the upper left of the central partial region 80. The dividing unit 41 also sets an intermediate partial region 80U between the central partial region 80 and the partial region 80 adjacent to the upper right of the central partial region 80. The dividing unit 41 also sets an intermediate partial region 80UR between the central partial region 80 and the partial region 80 adjacent to the upper right of the central partial region 80. The dividing unit 41 also sets an intermediate partial region 80L between the central partial region 80 and the partial region 80 adjacent to the left of the central partial region 80. The dividing unit 41 also sets an intermediate partial region 80R between the central partial region 80 and the partial region 80 adjacent to the right of the central partial region 80. The dividing unit 41 also sets an intermediate partial region 80DL between the central partial region 80 and the partial region 80 adjacent to the lower left of the central partial region 80. The dividing unit 41 also sets an intermediate partial region 80D between the central partial region 80 and the partial region 80 adjacent to the lower right of the central partial region 80. The dividing unit 41 also sets an intermediate partial region 80DR between the central partial region 80 and the partial region 80 adjacent to the lower right of the central partial region 80. Note that the overlap width when multiple partial regions 80 are overlapped is not limited to half the size of the partial regions 80, and may be any number of pixels.

[0074] In step SP35, the calculation unit 42 calculates a partial approximation line for each of the plurality of partial regions 80 that partially overlap with adjacent partial regions 80. In this case, if the calculation unit 42 can use the results of calculations that have already been performed for the adjacent partial regions 80, the calculation unit 42 may use the results of those calculations. This reduces the calculation load on the calculation unit 42.

[0075] According to this modification, the accuracy of fitting for the data point group 70 can be further improved.

[0076] (Second Modification) The dividing unit 41 may divide the entire region of the image 50 into a plurality of partial regions 80 of an optimal size, rather than a predetermined size, by updating the size of the partial regions 80. Furthermore, the dividing unit 41 may repeat updating the size in the reduction direction until the accuracy of fitting for the data point cloud 70 reaches or exceeds an allowable value.

[0077] FIG. 13 is a flowchart showing the processing contents executed by the processing unit 12, and shows a subroutine of step SP3 shown in FIG.

[0078] After the overall approximation line is generated by combining the partial approximation lines in step SP36, the calculation unit 42 then calculates the fitting accuracy of the overall approximation line generated in step SP36 in step SP32. As in the above, the fitting accuracy can be measured using, for example, the sum of squared residuals.

[0079] Next, in step SP33, the dividing unit 41 determines whether the accuracy calculated in step SP32 satisfies the tolerance condition (that is, whether the accuracy is equal to or greater than the above tolerance value).

[0080] If the accuracy satisfies the tolerance condition (that is, if the accuracy is equal to or greater than the tolerance value; step SP33: YES), the process returns to the main routine.

[0081] If the accuracy does not satisfy the tolerance condition (i.e., if the accuracy is less than the tolerance value; step SP33: NO), then in step SP34, the division unit 41 divides the image 50 into even finer regions. Thereafter, the processing from step SP35 onwards is repeatedly executed.

[0082] FIG. 14 is a diagram showing a schematic diagram of a state in which region division progresses in the reduction direction.

[0083] In (A) of Fig. 14, the division unit 41 divides the image 50 into three parts in the X direction and two parts in the Y direction. If, at this stage, the fitting accuracy of the overall approximation line generated by combining the partial approximation lines does not satisfy the tolerance condition, the division unit 41 divides the image 50 into six parts in the X direction and four parts in the Y direction, as shown in (B) of Fig. 14. If, at this stage, the fitting accuracy of the overall approximation line generated by combining the partial approximation lines does not satisfy the tolerance condition, the division unit 41 divides the image 50 into 12 parts in the X direction and 8 parts in the Y direction, as shown in (C) of Fig. 14. The division unit 41 repeatedly performs the same process until the fitting accuracy satisfies the tolerance condition.

[0084] According to this modification, the accuracy of fitting for the data point group 70 can be improved to an acceptable level or higher.

[0085] The present disclosure is particularly useful in application to object detection, which detects curved or straight line objects in an image.

Claims

1. An information processing method in which an information processing device acquires an image, generates a data point cloud corresponding to an object included in the image, calculates an approximation line by fitting the data point cloud, and in calculating the approximation line, divides the entire area of ​​the image into a plurality of partial areas, and calculates a partial approximation line that is a part of the approximation line by fitting a plurality of data points included in the partial area of ​​the data point cloud.

2. The information processing method according to claim 1, wherein the fitting includes nonlinear regression analysis, and the approximation line includes a regression curve.

3. The information processing method according to claim 1, wherein in generating the data point group, the data point group corresponding to the shape of the object is generated by instance segmentation of the image.

4. The information processing method according to claim 1, wherein the object includes a curve or a straight line contained in the image.

5. The information processing method according to claim 1, wherein in calculating the approximation line, the accuracy of fitting for the data point group is calculated, and if the accuracy is less than an allowable value, the region division and calculation of the partial approximation line are performed.

6. The information processing method according to claim 1, wherein the region division comprises dividing the entire region of the image into the plurality of partial regions of a predetermined size.

7. The information processing method according to claim 6, wherein the image has orthogonal first and second coordinates, and when the data point group includes a first data point and a second data point that have the same coordinate value of the first coordinate and whose coordinate value of the second coordinate differs by more than a threshold value, the predetermined size is set to a size that includes the first data point and the second data point in different partial areas.

8. The information processing method according to claim 6, wherein the shape of the partial region is set based on the aspect ratio of the image or the distribution shape of the data point group.

9. The information processing method according to claim 1, wherein in the region division, the entire region of the image is divided into the plurality of partial regions of optimal size by updating the sizes of the partial regions.

10. The information processing method according to claim 9, wherein the size update is repeated in the direction of reduction until the accuracy of fitting for the data point group is equal to or greater than an allowable value, and the optimal size is the size of the partial region at which the accuracy is equal to or greater than the allowable value.

11. The information processing method according to claim 9, wherein the shape of the partial region is set based on the aspect ratio of the image or the distribution shape of the data point group.

12. The information processing method according to claim 1, wherein in the region division, the entire region of the image is divided into the plurality of partial regions by overlapping parts of adjacent partial regions.

13. An information processing device comprising an information processing circuit, which acquires an image, generates a data point cloud corresponding to an object included in the image, calculates an approximation line by fitting the data point cloud, and in calculating the approximation line, divides the entire area of ​​the image into a plurality of partial areas, and calculates a partial approximation line that is a part of the approximation line by fitting a plurality of data points included in the partial area of ​​the data point cloud.

14. A program for causing an information processing device to execute a process, the process comprising: acquiring an image; generating a data point group corresponding to an object included in the image; calculating an approximation line by fitting the data point group; and, in calculating the approximation line, dividing the entire area of ​​the image into a plurality of partial areas; and calculating a partial approximation line that is a part of the approximation line by fitting a plurality of data points included in the partial area of ​​the data point group.

Images