Program generation support device of inspection system and inspection system
The program generation support device automates imaging position optimization by comparing non-defective and defective image data, addressing inefficiencies in manual inspection methods and enhancing defect detection accuracy.
Patent Information
- Application Number
- JP2023074420
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-04-28
- Publication Date
- 2024-11-08
AI Technical Summary
Existing inspection methods require repetitive manual optimization of imaging positions for photographing objects from multiple viewpoints, which is cumbersome and inefficient.
A program generation support device that automates the optimization of imaging positions by comparing non-defective and defective image data to generate a photography program efficiently, using a system comprising an image data acquisition unit, storage unit, and program modification unit to extract and modify photographing conditions.
Eliminates the need for manual annotation and optimization steps, enabling efficient and accurate imaging position optimization for inspection systems, improving the accuracy and efficiency of defect detection.
Smart Images

Figure 2024158849000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to a program generation support device for an inspection system, and to an inspection system. [Background technology]
[0002] Conventionally, in an inspection using image data obtained by photographing an object to be inspected from a plurality of viewpoints, the viewpoints at which the image data are photographed are optimized. This will be specifically described below.
[0003] (1) First, collect samples of both good and bad products for the object to be inspected. These samples may be collected from products or may be produced as prototypes.
[0004] (2) Next, the defective part of the defective sample is annotated by surrounding it with paint, etc. This makes the defective part identifiable by the inspection device. Next, the photographing position where the annotated defective part can be photographed is instructed and stored in the inspection device.
[0005] (3) The non-defective samples and the defective samples are photographed by an image capture device such as a camera, and the image data is stored in the inspection device, thereby collecting image data of the non-defective samples and the defective samples.
[0006] (4) The worker verifies the collected image data and extracts inappropriate image data. For example, for defective product samples, image data in which the defective part cannot be identified because it is hidden by other parts, or the defective part is too small in the image data to be identified, etc., are extracted.
[0007] (5) The extracted image data is re-taught to the inspection device, and the shooting position of the image acquisition device is optimized. This optimization is performed by the operator individually for each image data.
[0008] By repeating the above-mentioned operations (2) to (5), the photographing position of the image data is optimized.
[0009] Inspection using image data obtained by photographing an object to be inspected is described in, for example, Patent Document 1. [Prior art documents] [Patent documents]
[0010] [Patent Document 1] JP 2020-187661 A Summary of the Invention [Problem to be solved by the invention]
[0011] In an inspection using image data obtained by photographing an object to be inspected from a plurality of viewpoints, it is extremely troublesome to repeatedly carry out the above-mentioned operations (2) to (5).
[0012] The present invention has been made in consideration of such problems, and aims to provide a program generation support device for an inspection system, and an inspection system, which are capable of creating a program that can efficiently optimize the shooting position when photographing an object to be inspected from multiple viewpoints. [Means for solving the problem]
[0013] One aspect of the present invention is A program generation support device for an inspection system that is applied to an inspection system that inspects an object to be inspected and supports generation of an imaging program for imaging the object to be inspected, comprising: an image data acquisition unit that acquires a plurality of image data by photographing the inspection object from a plurality of viewpoints; a storage unit that stores a base program, which is a program for photographing the plurality of image data under a base photographing condition, a plurality of non-defective image data, which are image data of the inspection objects labeled as non-defective products, and a plurality of defective image data, which are image data of the inspection objects labeled as defective products; a difference image data extraction unit that compares the non-defective image data with the defective image data captured from the same viewpoint as the non-defective image data, and extracts difference image data from the defective image data that differs from the non-defective image data; The program generation support device for an inspection system includes a program correction unit that extracts the shooting conditions under which the difference image data was captured from the base shooting conditions, and generates the shooting program based on the shooting conditions.
[0014] Another aspect of the present invention is The inspection system includes an inspection program for photographing the inspection object based on the photographing program generated by the program generation support device of the inspection system. Effect of the Invention
[0015] According to one aspect of the present invention, it is possible to compare non-defective image data and defective image data photographed from the same viewpoint, extract the photographing conditions under which the different image data was photographed from the base photographing conditions, and generate a photographing program based on the extracted photographing conditions. This eliminates the need for an operator to repeat the steps of annotating defective parts, teaching the photographing position, photographing, and verifying, and therefore makes it possible to efficiently optimize the photographing position of the inspection object.
[0016] According to another aspect of the present invention, an inspection object can be inspected based on an imaging program in which the imaging position of the inspection object is efficiently optimized. [Brief description of the drawings]
[0017] [Figure 1] FIG. 1 is a block diagram showing a program generation support device and an inspection system according to a first embodiment. [Diagram 2] FIG. 2 is a partially cutaway front view showing the imaging device of the first embodiment. [Diagram 3] 4 is a flowchart showing the operation of the program generation support device and the inspection system according to the first embodiment. [Figure 4]6 is a table for explaining a process of extracting shooting conditions from base shooting conditions according to the first embodiment. [Diagram 5] 5A to 5C are diagrams for explaining a process of comparing non-defective image data with defective image data in the image data comparison process according to the first embodiment. [Figure 6] 5A to 5C are diagrams showing display screens displayed in the difference image data display process according to the first embodiment. [Figure 7] FIG. 11 is a diagram showing a state in which one different image data is selected in the selection process of the first embodiment. [Figure 8] FIG. 11 is a block diagram showing a program generation support device and a test system according to a second embodiment. [Figure 9] 10 is a flowchart showing the operation of a program generation support device and a testing system according to a second embodiment. [Figure 10] 13A to 13C are diagrams showing display screens displayed in the difference image data display process according to the second embodiment. [Figure 11] FIG. 11 is a partially cutaway front view showing an imaging device according to another embodiment (3). DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0018] (Embodiment 1) 1. Configuration of program generation support device 10 and inspection system 11 1 and 2, a first embodiment in which a program generation support device 10 of the present invention is applied to an inspection system 11 will be described. The inspection system 11 of this embodiment inspects an inspection object 12 (see FIG. 2). The program generation support device 10 of this embodiment also supports the generation of a photography program SP for photographing the inspection object 12. The inspection object 12 of this embodiment is not particularly limited, but can be, for example, an automobile part.
[0019] The inspection system 11 and the program generation support device 10 include an imaging device 13, a control device 14, a storage unit 15, a display device 16, an input device 17, a learning model generation unit 18, a difference image data extraction unit 19, and a program correction unit 20. The imaging device 13 includes a camera 13a (an example of an image data acquisition unit), a lighting device 13b, a servo motor 13c, and a mounting table 13d.
[0020] As shown in FIG. 2, the photographing device 13 includes a base 21, a cover 22 attached to the upper side of the base 21 and covering the upper surface of the base 21, a mounting stand 13d arranged on the upper surface of the base 21, a robot arm 13e arranged on the upper surface of the base 21 at a position different from the mounting stand 13d, a camera 13a arranged at the tip of the robot arm 13e, and a lighting device 13b attached to the camera 13a.
[0021] The inspection target object 12 is placed on the upper surface of the mounting table 13d. The mounting table 13d is configured to be rotatable about an axis in the vertical direction. The rotation angle θ of the mounting table 13d is controlled by a servo motor 13c built in the mounting table 13d. The range of the rotation angle θ of the mounting table 13d is not particularly limited and is set to an arbitrary range. In this embodiment, it is set to, for example, 0° to 360°.
[0022] The robot arm 13e comprises a base 23, a first joint portion 24 arranged on the upper part of the base 23, a first arm portion 25 extending from the first joint portion 24, a second joint portion 26 arranged at the tip of the first arm portion 25, a second arm portion 27 extending from the second joint portion 26, a third joint portion 28 arranged at the tip of the second arm portion 27, and a third arm portion 29 extending from the third joint portion 28.
[0023] The base 23 is configured to be rotatable about an axis in the vertical direction. The rotation angle γ of the base 23 is controlled by a servo motor 13c built in the base 23. The range of the rotation angle γ of the base 23 is not particularly limited and can be set to an arbitrary range. In this embodiment, the rotation angle γ is set to, for example, 0° to 90° so that the robot arm 13e can scan the inspection target 12 placed on the placement table 13d.
[0024] The second arm section 27 is configured such that a portion on the second joint section 26 side and a portion on the third joint section 28 side are relatively rotatable. The rotation angle φ of the second arm section 27 is controlled by a servo motor 13c built in the second arm section 27. The range of the rotation angle φ of the second arm section 27 is not particularly limited and can be set to any range. In this embodiment, the range is set to, for example, 0° to 180° so that a surface of the inspection object 12 placed on the placement table 13d other than the surface facing the placement table 13d can be scanned.
[0025] The first joint portion 24, the second joint portion 26, and the third joint portion 28 are configured to be bendable. The bending angles of the first joint portion 24, the second joint portion 26, and the third joint portion 28 are controlled by servo motors 13c built into the first joint portion 24, the second joint portion 26, and the third joint portion 28, respectively.
[0026] Camera 13a is configured to be movable in the X-axis direction, the Y-axis direction, and the Z-axis direction by robot arm 13e. In addition, mounting table 13d is configured to be rotatable about an axis in the vertical direction, so that the inspection target 12 placed on mounting table 13d can be photographed by camera 13a in a portion of inspection target 12 excluding the surface of inspection target 12 facing mounting table 13d.
[0027] The cover 22 covering the upper surface of the base 21 can block external light when photographing the inspection object 12. In addition, when photographing the inspection object 12, it is possible to prevent an object other than the inspection object 12 from being captured in the image data. The lighting device 13b attached to the camera 13a irradiates the inspection object 12 with light. The lighting device 13b is attached to the tip of an attachment member (not shown) extending from the camera 13a. The lighting device 13b can be any light source selected as appropriate, such as a fluorescent lamp, a cold cathode fluorescent lamp, an LED, etc. The camera 13a can be any imaging device 13 selected as appropriate, such as a CCD camera, an optical camera, etc.
[0028] As shown in FIG. 1, the storage unit 15 stores non-defective image data GD, defective image data BD, a base program BP, a photography program SP, base photography conditions BC, photography conditions SC, a non-defective learning model GM, and difference image data DD.
[0029] The good product image data GD is image data obtained by photographing an inspection object 12 that does not have a defective portion 30. The defective product image data BD is image data obtained by photographing an inspection object 12 that has a defective portion 30. Some of the defective product image data BD contains defective portions 30, depending on the viewpoint from which the image was taken.
[0030] The base program BP is a program for photographing the inspection object 12 by the imaging device 13 based on the base imaging conditions BC. The base imaging conditions BC are a database of imaging conditions SC when the inspection object 12 is photographed by the imaging device 13. The base imaging conditions BC include the position and posture of the camera 13a, the brightness of the lighting device 13b, etc.
[0031] In the base imaging condition BC, the movement amounts for the X-axis, Y-axis, and Z-axis are set in as fine a division as possible when imaging the inspection object 12, and the movement amounts for the rotation angles θ, φ, and γ are also set in as fine a division as possible. That is, the base program BP includes an extremely large amount of information or commands when imaging the inspection object 12.
[0032] The photographing program SP is a program generated by the program modification unit 20. The photographing program SP is a program for photographing the inspection object 12 by the photographing device 13 based on the photographing conditions SC. The photographing conditions SC are generated by the program modification unit 20 by modifying the base photographing conditions BC.
[0033] The shooting conditions SC are generated by extracting a part of the base shooting conditions BC. That is, the shooting conditions SC are composed of a part of a great number of commands or information included in the base shooting conditions BC. Furthermore, the shooting conditions SC may be generated by modifying the commands and information extracted from the base shooting conditions BC.
[0034] The good product learning model GM is formed by the learning model generation unit 18 through machine learning of the good product image data GD.
[0035] The difference image data DD is formed by extracting images different from the good image data GD from the defective image data BD by the difference image data extraction unit 19. In detail, the good image data GD is compared with the defective image data BD captured from the same viewpoint as the good image data GD, and the defective image data BD that differs from the good image data GD is extracted as the difference image data DD.
[0036] The learning model generation unit 18 generates a good product learning model GM by machine learning the good product image data GD stored in the storage unit 15, and stores the good product learning model GM in the storage unit 15.
[0037] The different image data extraction unit 19 compares the good-quality image data GD with the defective-quality image data BD captured from the same viewpoint as the good-quality image data GD, extracts the defective-quality image data BD that differs from the good-quality image data GD as different image data DD, and stores the different image data DD in the storage unit 15. The same viewpoint means that the position and attitude of the camera 13a are the same.
[0038] The difference image data extraction unit 19 calculates the Euclidean distance between the non-defective learning model GM and the difference image data DD.
[0039] The program modification unit 20 extracts the shooting conditions SC under which the difference image data DD was shot from the base shooting conditions BC, and generates the shooting program SP based on the shooting conditions SC. The program modification unit 20 may generate the shooting program SP by further modifying the shooting conditions SC extracted from the base shooting conditions BC.
[0040] The correction made by the program correction unit 20 may be, for example, to reduce the distance between the inspection object 12 and the camera 13a. This allows the inspection object 12 to be photographed in an enlarged scale, so that the defective part 30 of the defective product can be photographed clearly.
[0041] In addition, the program correction unit 20 may correct the relative photographing position between the inspection object 12 and the camera 13a. When photographing based on the base photographing condition BC, it is effective when the defective portion 30 is hidden by other members and cannot be visually recognized.
[0042] Moreover, the program correcting unit 20 may correct the brightness of the lighting device 13b. This is effective when the defective portion 30 is not sufficiently illuminated with light when photographing based on the base photographing condition BC.
[0043] The display device 16 displays the difference image data DD. The display device 16 displays the difference image data DD in descending order of the Euclidean distance calculated by the difference image data extraction unit 19. This allows the worker to visually recognize the difference image data DD displayed on the display device 16 in descending order of the difference from the good product learning model GM.
[0044] The operator selects the optimum image data OD used to generate the photographing program SP from the difference image data DD displayed on the display device 16, and inputs it using the input device 17. This makes it possible to generate the photographing program SP by correcting the base program BP based on the conditions under which the optimum image data OD was photographed. As a result, the accuracy of the generated program can be improved.
[0045] 2. Operation of the program generation support device 10 3 to 7, the operations of the program generation support device 10 and the inspection system 11 of the present embodiment will be described. However, the operations of the program generation support device 10 and the inspection system 11 are not limited to the following description.
[0046] 3, when the program generation support device 10 and the inspection system 11 are started, a non-defective product photographing process (S1) is executed. In the non-defective product photographing process (S1), a plurality of inspection objects 12 that do not have a defect 30 are collected, and each inspection object 12 is photographed by the photographing device 13.
[0047] First, the inspection object 12 is placed on the placement table 13d. Next, the inspection object 12 is photographed according to the base program BP based on the base photographing condition BC. That is, the first joint 24, the second joint 26, the third joint 28, the placement table 13d, and the second arm 27 of the robot arm 13e are driven by the servo motor 13c, so that the camera 13a moves in the X-axis direction, the Y-axis direction, and the Z-axis direction, while the placement table 13d is rotated by the servo motor 13c at a rotation angle θ. The amount of movement in the X-axis direction, the Y-axis direction, and the Z-axis direction is not particularly limited and can be set to any value. In this embodiment, for example, the amount of movement in the X-axis direction, the Y-axis direction, and the Z-axis direction is set to 300 mm. Also, as described above, the rotation angle θ is set to 0° to 360°, the rotation angle φ is set to 0° to 180°, and the rotation angle γ is set to 0° to 90°.
[0048] The feed amounts in the X-axis, Y-axis, and Z-axis directions are not particularly limited and are set to any value. In the base photographing condition BC of this embodiment, the feed amounts in the X-axis, Y-axis, and Z-axis are set to, for example, 10 mm. In addition, the feed amounts of the rotation angles θ, φ, and γ are not particularly limited and are set to any value. In the base photographing condition BC of this embodiment, the feed amounts of the rotation angles θ, φ, and γ are set to, for example, 10°.
[0049] Next, a non-defective product learning model generation process (S2) is executed. In the non-defective product learning model generation process (S2), the learning model generation unit 18 performs machine learning based on a plurality of non-defective product image data GD to generate a non-defective product learning model GM.
[0050] Next, the defective product photographing process (S3) is executed. In the defective product photographing process (S3), the inspection objects 12 are photographed according to the base program BP based on the base photographing conditions BC for defective products having defective parts 30. The base program BP is the same as the non-defective product photographing process (S1), so a duplicated description will be omitted.
[0051] In Table 1 of FIG. 4, the base photographing conditions BC photographed by the defective product photographing process are listed with some parts omitted. The NO. on the left side is the number of defective product image data BD. In this embodiment, for example, N defective product image data BD are photographed. X, Y, and Z listed in the upper row are the feed amounts in the X-axis direction, Y-axis direction, and Z-axis direction. X, Y, and Z are fed in 10 mm increments from 0 mm to 300 mm. Also, θ, φ, and γ listed in the upper row are the feed amounts of the rotation angle θ, the rotation angle φ, and the rotation angle γ. The rotation angle θ is fed from 0° to 360°, the rotation angle φ is fed from 0° to 180°, and the rotation angle γ is fed from 0° to 90° in 10° increments.
[0052] Next, returning to FIG. 3, an image data comparison process (S4) is executed. In the image data comparison process (S4), the difference image data extraction unit 19 compares the good-quality image data GD with the defective-quality image data BD captured from the same viewpoint as the good-quality image data GD. If the good-quality image data GD and the defective-quality image data BD differ, the difference image data extraction unit 19 extracts the different defective-quality image data BD. On the other hand, if the good-quality image data GD and the defective-quality image data BD do not differ, the difference image data extraction unit 19 does not extract anything. This point will be described with reference to FIG. 5(a) to FIG. 5(c).
[0053] 5(a), when the good image data GD and the defective image data BD are compared, a defective portion 30 is shown in the defective image data BD. The good image data GD and the defective image data BD differ at this defective portion 30. Therefore, the difference image data extraction unit 19 extracts this defective image data BD as difference image data DD.
[0054] 5(b), when comparing the good image data GD and the defective image data BD, the defective part 30 is not shown in the defective image data BD from this viewpoint. Therefore, from this viewpoint, the good image data GD and the defective image data BD do not differ. Therefore, the difference image data extraction unit 19 does not extract the defective image data BD.
[0055] 5(c) shows good product image data GD and defective product image data BD, which are images of an enlarged portion of the inspection target 12. From this viewpoint, the defective product image data BD does not show the defective portion 30. Therefore, from this viewpoint, there is no difference between the good product image data GD and the defective product image data BD. Therefore, the difference image data extraction unit 19 does not extract the defective product image data BD.
[0056] 3, a difference determination process (S5) is executed. In the difference determination process (S5), the difference image data extraction unit 19 calculates the Euclidean distance between the non-defective learning model GM and the extracted difference image data DD.
[0057] Next, in a difference image data display process (S6), the display device 16 displays the difference image data DD in descending order of Euclidean distance. FIG. 6 shows a display screen 31 displayed on the display device 16. Difference image data DD labeled 1, 2, and 3 are displayed on the display screen 31. The difference image data DD labeled 1 has the largest Euclidean distance from the good product learning model GM, the difference image data DD labeled 2 has the second largest Euclidean distance, and the difference image data DD labeled 3 has the smallest Euclidean distance. However, the number of difference images displayed on the display screen 31 is not particularly limited.
[0058] Next, returning to FIG. 3, a selection process (S7) is executed. In the selection process (S7), the operator selects one or more optimum image data OD from the plurality of difference image data DD displayed on the display device 16. As shown in FIG. 7, a check box displayed below the label 1 is checked, thereby indicating that the difference image data DD labeled with the label 1 has been selected as the optimum image data OD. The operator uses the input device 17, such as a mouse or a touch panel, to check a check box displayed to the side of the difference image data DD labeled with the label 1, thereby selecting the difference image data DD labeled with the label 1 as the optimum image data OD. The method of selecting the optimum image data OD on the display screen 31 is arbitrary and is not limited to a check box.
[0059] Next, returning to FIG. 3, the base program correction process (S8) is executed. In the base program correction process (S8), the program correction unit 20 extracts, from the base shooting conditions BC, the shooting conditions SC under which the optimum image data OD extracted from the difference image data DD was shot. In Table 1 of FIG. 4, the numbers enclosed by dashed lines indicate the defective image data BD selected as the optimum image data OD. Table 2 of FIG. 4 shows the shooting conditions SC of the selected optimum image data OD. In Table 2, three sets of shooting conditions SC are shown, but the shooting conditions SC of the selected optimum image data OD may be one or two, or four or more.
[0060] As shown in Tables 1 and 2 in Figure 4, by extracting from the base shooting conditions BC shown in Table 1, the number of shooting conditions SC of the optimal image data OD shown in Table 2 is less than the number of base shooting conditions BC in Table 1.
[0061] Based on the shooting conditions SC of the optimum image data OD, the program correction unit 20 generates a shooting program SP for shooting the inspection object 12. With the above, the operation of the program generation support device 10 is completed.
[0062] 3, the inspection system 11 executes an inspection process (S9) of the inspection object 12 based on the photography program SP generated by the program generation support device 10. With the above, the operation of the inspection system 11 ends.
[0063] 3. Effects of this Form Next, the effects of the present embodiment will be described. The program generation support device 10 for the inspection system 11 according to the present embodiment is applied to the inspection system 11 that inspects the inspection object 12, and supports the generation of a photography program SP that photographs the inspection object 12. The program generation support device 10 includes a camera 13a, a storage unit 15, a difference image data extraction unit 19, and a program correction unit 20.
[0064] The camera 13a captures a plurality of image data by photographing the inspection object 12 from a plurality of viewpoints. The storage unit 15 stores a base program BP, which is a program for photographing a plurality of image data under a base photographing condition BC, a plurality of good product image data GD, which is image data of the inspection object 12 labeled as a good product, and a plurality of defective product image data BD, which is image data of the inspection object 12 labeled as a defective product. The difference image data extraction unit 19 compares the good product image data GD with the defective product image data BD photographed from the same viewpoint as the good product image data GD, and extracts the difference image data DD, which differs from the good product image data GD, from the defective product image data BD. The program correction unit 20 extracts the photographing condition SC under which the difference image data DD was photographed from the base photographing condition BC, and generates a photographing program SP based on the photographing condition SC.
[0065] According to this embodiment, since the operator does not need to repeat the steps of annotating the defective portion 30, teaching the photographing position, photographing, and verification, the photographing position of the inspection target 12 can be efficiently optimized.
[0066] The program generation support device 10 according to this embodiment includes a learning model generation unit 18 that generates a good product learning model GM based on a plurality of good product image data GD. The difference image data extraction unit 19 extracts difference image data DD from the defective product image data BD based on the good product learning model GM generated by the learning model generation unit 18. This allows the difference image data DD to be extracted based on the good product learning model GM, thereby improving the accuracy of the difference image data DD. As a result, the accuracy of the inspection performed based on the shooting program SP can be improved.
[0067] According to this embodiment, the program correction unit 20 generates the imaging program SP by further correcting the imaging conditions SC. By further correcting the imaging conditions SC extracted from the base imaging conditions BC, the accuracy of the examination performed based on the imaging program SP can be further improved.
[0068] According to this embodiment, the program correction unit 20 generates the photographing program SP by making a correction to reduce the distance between the inspection object 12 and the camera 13a. This allows the defective part 30 to be photographed in an enlarged manner, which is effective when the defective part 30 is small.
[0069] According to this embodiment, the program correction unit 20 generates the photography program SP by correcting the relative photography positions of the inspection target object 12 and the camera 13a. This is effective when the defective portion 30 is hidden by other members when photographed based on the base photography condition BC.
[0070] The program generation support device 10 according to this embodiment includes an illumination device 13b that irradiates light onto an inspection target 12. The program correction unit 20 generates a photography program SP by correcting the brightness of the illumination device 13b. This is effective when sufficient light is not irradiated onto a defective portion 30 when the image is captured based on the base photography condition BC.
[0071] The program generation support device 10 according to this embodiment includes an input device 17 for selecting optimal image data OD used to generate a shooting program SP from the difference image data DD displayed on a display device 16. A program correction unit 20 corrects the base program BP based on the conditions under which the selected optimal image data OD was photographed. This allows the operator to select the optimal image data OD from the displayed difference image data DD, thereby reducing the labor required for generating the shooting program SP.
[0072] (Embodiment 2) Next, a second embodiment will be described with reference to Fig. 8 to Fig. 10. As shown in Fig. 8, an inspection system 11a and a program generation support device 10a according to this embodiment differ from those of the first embodiment in that they do not include an input device 17.
[0073] Fig. 9 shows a flowchart showing the operation of the inspection system 11a and the program generation support device 10a according to this embodiment. This embodiment differs from the first embodiment in that the selection process is omitted. Therefore, as shown in Fig. 10, in the difference image data display process (S16), the check box for selecting the optimum image data OD is omitted from the display screen 31 displayed on the display device 16.
[0074] Display screen 31 displays difference image data DD labeled 1, 2, and 3. The meanings of the label numbers are the same as in embodiment 1, so duplicated explanations will be omitted. An operator who views display screen 31 can confirm what image data the program was generated based on.
[0075] In addition, in this embodiment, since the selection process is omitted, in the base program correction process (S18), the program correction unit 20 extracts the conditions under which the difference image data DD was captured from the base shooting conditions BC, and generates the shooting program SP for capturing images of the inspection object 12 based on the shooting conditions SC. According to this embodiment, the shooting program SP for capturing images of the inspection object 12 can be generated without the operator having to select image data, so that the number of worker's work steps can be reduced.
[0076] In addition, among the symbols used in the second and subsequent embodiments, the same symbols as those used in the previous embodiments represent the same components, etc. as those in the previous embodiments, unless otherwise specified.
[0077] The present invention is not limited to the above-described embodiments, and can be applied to various embodiments without departing from the gist of the present invention.
[0078] (1) In the first and second embodiments, the control device 14 may be configured to be included in the photographing device 13.
[0079] (2) In the first and second embodiments, the learning model generation unit 18 may be omitted.
[0080] (3) As shown in Fig. 11, a rotatable revolver 32 may be attached to the camera 13a, and a plurality of lenses 33 with different magnifications may be attached to the revolver 32. The rotation angle of the revolver 32 is controlled by the control device 14. By rotating the revolver 32 to a predetermined position, the camera 13a can capture images of the inspection object 12 with different magnifications. [Explanation of symbols]
[0081] 10, 10a: program generation support device, 11, 11a: inspection system, 12: inspection object, 13a: camera, 13b: lighting device, 15: memory unit, 16: display device, 17: input device, 18: learning model generation unit, 19: difference image data extraction unit, 20: program correction unit, 30: defective area, BC: base shooting conditions, BD: defective product image data, BP: base program, DD: difference image data, GD: good product image data, GM: good product learning model, OD: optimal image data, SC: shooting conditions, SP: shooting program
Claims
1. A program generation support device for an inspection system that is applied to an inspection system that inspects an object to be inspected and supports generation of an imaging program for imaging the object to be inspected, comprising: an image data acquisition unit that acquires a plurality of image data by photographing the inspection object from a plurality of viewpoints; a storage unit that stores a base program, which is a program for photographing the plurality of image data under a base photographing condition, a plurality of non-defective image data, which are image data of the inspection objects labeled as non-defective products, and a plurality of defective image data, which are image data of the inspection objects labeled as defective products; a difference image data extraction unit that compares the non-defective image data with the defective image data captured from the same viewpoint as the non-defective image data, and extracts difference image data from the defective image data that differs from the non-defective image data; a program correction unit that extracts the shooting conditions under which the difference image data was shot from the base shooting conditions, and generates the shooting program based on the shooting conditions.
2. moreover, a learning model generation unit that generates a good product learning model based on the plurality of good product image data, 2. The program generation support device for an inspection system according to claim 1, wherein the difference image data extraction unit extracts the difference image data from the defective product image data based on the non-defective product learning model generated by the learning model generation unit.
3. moreover, a display device for displaying the difference image data; The difference image data extraction unit calculates a Euclidean distance between the good product learning model and the difference image data, 3. The program generation support device for an inspection system according to claim 2, wherein the display device displays the difference image data in descending order of the Euclidean distance.
4. The program correction unit 2. The program generation support device for an inspection system according to claim 1, further comprising: a step of generating the imaging program by further modifying the imaging conditions.
5. 5. The program generation support device for an inspection system according to claim 4, wherein the program correction unit generates the imaging program by making a correction to reduce a distance between the inspection object and the image data acquisition unit.
6. 5. The program generation support device for an inspection system according to claim 4, wherein the program correction unit generates the imaging program by correcting a relative imaging position between the inspection object and the image data acquisition unit.
7. moreover, A lighting device that irradiates the inspection object with light, The program generation support device for an inspection system according to claim 4 , wherein the program correction unit generates the photography program by correcting brightness of the lighting device.
8. moreover, a display device for displaying the difference image data; an input device for selecting optimal image data to be used for generating the photography program from the difference image data displayed on the display device; 2. The program generation support device for an inspection system according to claim 1, wherein the program correction section corrects the base program based on conditions under which the selected optimum image data was captured.
9. moreover, a learning model generation unit that generates a good product learning model based on the plurality of good product image data, The difference image data extraction unit calculates a Euclidean distance between the good product learning model and the difference image data, 9. The program generation support device for an inspection system according to claim 8, wherein the display device displays the difference image data in descending order of the Euclidean distance.
10. 2. The program generation support device for an inspection system according to claim 1, wherein the object to be inspected is an automobile part.
11. An inspection system that causes an inspection object to be photographed based on the photography program generated by the program generation support device of the inspection system according to any one of claims 1 to 10.
Citation Information
Patent Citations
Image inspection device
JP2020187661A
