JP2010281786A - Evaluation system of visual inspection apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an evaluation system of a visual inspection apparatus for evaluating the work of a wide range of negative instances without acquiring any actual work of positive and negative instances. <P>SOLUTION: A negative instance image compositing section 24 sticks a two-dimensional faulty section image generated by a two-dimensional faulty section image generation section 23 to a pseudo three-dimensional image of work generated artificially by a three-dimensional image generation section 22 for compositing a three-dimensional negative instance image. An inspection apparatus 13 evaluates the visual inspection apparatus by a pseudo two-dimensional image obtained by artificially photographing the composited three-dimensional negative instance image. By accumulating a number of faulty sections such as scratches and discoloration, as a two-dimensional faulty section image, a two-dimensional faulty section image is simply stuck to a pseudo three-dimensional image of artificial work, and thereby a three-dimensional negative instance image corresponding to the work of the negative instance is created easily. By utilizing the three-dimensional negative instance image, the visual inspection apparatus is evaluated for the work of a wide range of the negative instances. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an evaluation system for a visual inspection apparatus.

  2. Description of the Related Art Conventionally, a visual inspection apparatus is known in which, for example, a workpiece to be inspected is photographed by a camera attached to the tip of a robot arm or a camera fixed to an inspection facility, and the workpiece is inspected using the photographed image. Such a visual inspection apparatus needs to adjust the evaluation of the inspection result and adjust the judgment criteria in advance so that an appropriate inspection result can be obtained in the inspection facility at the time of actual use. For example, in the case of Patent Document 1, a visual inspection apparatus is evaluated by installing a work in actual inspection equipment and photographing a real work. In the case of Patent Document 1, a workpiece that passes the inspection among real workpieces, that is, a so-called positive example workpiece is photographed, and the visual inspection device is evaluated using the photographed image.

  On the other hand, in order to increase the inspection accuracy of the visual inspection program and the visual inspection program for inspecting the workpiece using this, not only the positive example work but also the work that fails the inspection, so-called negative case work, It is necessary to evaluate how the visual inspection device and the visual inspection program behave. However, the negative example work needs to be created by adding, for example, a defective part that causes an inspection failure such as a scratch or discoloration to the positive example work. Therefore, there is a problem that a defective part must be added to the work of the positive case only for evaluation of the visual inspection device and the visual inspection program, which is not realistic. In addition, there is a problem that it is difficult to obtain work of negative cases corresponding to all of the assumed negative cases.

JP 2002-213929 A

  SUMMARY OF THE INVENTION An object of the present invention is to provide an evaluation system for a visual inspection apparatus that evaluates a wide range of negative case workpieces without obtaining real positive and negative case workpieces.

  The present inventor has found that when an image acquired by photographing with a camera is image-processed for inspection, an image actually obtained by photographing with a camera is substantially the same as a negative example image created in a pseudo manner. It was. According to the first aspect of the present invention, the 2D defective partial image generated by the 2D (two-dimensional) defective partial image generating unit is pasted on the 3D image of the workpiece that is artificially generated by the 3D (three-dimensional) image generating unit. . Even when an actual work is used, the visual inspection apparatus is finally evaluated by an image obtained by photographing the work, that is, a 2D image. For this reason, even if a 2D defective partial image is pasted on a pseudo 3D image of a workpiece, the resulting image is a pseudo 2D image that is substantially the same as an image of the actual workpiece. Further, the 2D defective part image representing the defective part can be arbitrarily expressed in consideration of, for example, the imaging direction of the workpiece in the visual inspection apparatus. When a pseudo 2D image is used for the evaluation of the visual inspection device, an image actually captured by the camera is also a 2D image having no depth. As a result, even if the 2D defective part image is pasted on the 3D image of the work, a pseudo 2D image equivalent to that obtained by photographing the work having the 3D defective part is created. Thus, by accumulating many defective parts such as scratches and discoloration as 2D defective part images, it is possible to easily attach a 2D defective part image to a 3D image of a pseudo work and easily correspond to a negative example work. A 3D negative case image is created. By using this 3D negative case image, the visual inspection apparatus is evaluated for a wide range of negative case works. Therefore, it is possible to evaluate a wide range of negative case works in the visual inspection apparatus without obtaining real cases of positive cases and negative cases, and to improve the inspection accuracy of the visual inspection apparatus.

The block diagram which shows the evaluation system of the visual inspection apparatus by one Embodiment of this invention. The schematic diagram which shows the structure of the test | inspection data file of the evaluation system by one Embodiment of this invention. Schematic which shows the flow of a process of the evaluation system by one Embodiment of this invention. The schematic diagram which shows the data flow of the evaluation system by one Embodiment of this invention. The schematic diagram which shows the 2D defect partial image of the evaluation system by one Embodiment of this invention. Explanatory drawing which shows the relationship between the real image and pseudo image which are used in order to verify the evaluation system by one Embodiment of this invention Schematic diagram showing the flow of binary difference image extraction for verifying the evaluation system Schematic diagram showing the flow from the real image of the positive case and negative case to the generation of the binary difference image Schematic diagram showing the flow from the generation of positive and negative pseudo images to the generation of pseudo binary difference images Schematic showing an example of a visual inspection device

Hereinafter, an embodiment of an evaluation system for a visual inspection apparatus according to the present invention will be described with reference to the drawings.
(Visual inspection device)
First, the visual inspection apparatus 100 shown in FIG. 10 will be described. The visual inspection apparatus 100 shown in FIG. 10 is substantial and has no direct relationship with the visual inspection apparatus evaluation system of the present embodiment, but will be described for the sake of easy understanding. The visual inspection device 100 includes an inspection facility 101, a camera 102, and a processing device 103. In the inspection facility 101, the workpiece 104 that is actually inspected is inspected. The camera 102 is provided in the inspection facility 101. The camera 102 is fixed to the inspection facility 101 or attached to a robot 105 provided in the inspection facility 101. When the camera 102 is fixed to the inspection facility 101, the workpiece 104 is always photographed from a certain position. On the other hand, when the camera 102 is attached to the robot 105, the work 104 is photographed from an arbitrary position around. Whether the camera 102 is fixed to the inspection facility 101 or attached to the robot 105 can be arbitrarily selected according to the type of inspection of the workpiece 104.

  The processing apparatus 103 is configured by, for example, a personal computer (hereinafter, “personal computer”) and the like, and inspects the workpiece 104 based on a photographed image of the workpiece 104 photographed by the camera 102. The processing device 103 inspects the workpiece 104 according to a preset visual inspection program. Specifically, the processing apparatus 103 acquires a planar image, that is, a 2D image of the workpiece 104 photographed by the camera 102 by executing a visual inspection program. In this case, the processing apparatus 103 acquires an image of the actual workpiece 104 captured by the camera 102, that is, a real image of the workpiece 104. Then, the processing apparatus 103 binarizes the acquired 2D image of the workpiece 104, and inspects the workpiece 104 based on the binarized image. Here, the inspection items include, for example, the presence or absence of scratches or discoloration of the workpiece 104, the dimensions of the entire workpiece 104 and each part, the shape of the workpiece 104, and the like.

The visual inspection apparatus 100 as described above actually photographs the workpiece 104 with the camera 102. Therefore, when evaluating the visual inspection device 100, it is necessary to actually construct the visual inspection device 100 and acquire a real 2D image using the real object of the workpiece 104. Among these, the workpieces that pass the inspection among the actual workpieces 104, that is, the so-called positive case workpieces and their images can be easily obtained by obtaining the workpieces distributed with the product. On the other hand, in order to increase the inspection accuracy of the visual inspection apparatus 100 and the visual inspection program for inspecting the workpiece 104 using the visual inspection device 100, not only positive case workpieces but also workpieces that fail inspection, so-called negative case workpieces. In addition, it is necessary to evaluate the behavior of the visual inspection device 100 and the visual inspection program. However, the negative example work needs to be created by adding, for example, a defective part that causes an inspection failure such as a scratch or discoloration to the positive example work. For this reason, a defective portion must be added to the work in the positive case only for the evaluation of the visual inspection device 100 and the visual inspection program, which is not realistic. It is also difficult to obtain negative case work for all possible negative cases.
Therefore, the visual inspection device evaluation system of the present embodiment described below is intended to evaluate the inspection result without acquiring an actual workpiece with the actual visual inspection device.

(Evaluation system according to this embodiment)
As shown in FIG. 1, the evaluation system 10 of this embodiment includes a control device 11, an image generation device 12, an inspection device 13, a storage device 14, and a display device 15. The evaluation system 10 artificially constructs a visual inspection device 100 as shown in FIG. 10 in a virtual world, and the virtual inspection device 100 thus constructed and a visual inspection program executed by the visual inspection device 100. Conduct an assessment of

  The evaluation system 10 is realized by software by executing an evaluation program in a personal computer, for example. Note that the evaluation system 10 may be realized not only by software but also by hardware. The control device 11 is composed of, for example, a microcomputer having a CPU, a ROM, a RAM, and the like, and controls the entire evaluation system 10 by executing the above-described visual inspection program and evaluation program.

  The image generation apparatus 12 includes a work data acquisition unit 21, a pseudo 3D image generation unit 22, a 2D defective partial image generation unit 23, a negative case image synthesis unit 24, a 2D camera image generation unit 25, a 2D image transmission unit 26, and a storage data transmission. A portion 27 is provided. The work data acquisition unit 21 acquires work data serving as a basis for forming a work in a pseudo manner. The work data acquisition unit 21 acquires the work data input from the input unit 28. Various parameters constituting the work data are input to the input unit 28. The input unit 28 performs mechanical input using, for example, a keyboard or a mouse, and optical or magnetic input using a CD or HDD. The work data acquired by the work data acquisition unit 21 is data serving as a basis for artificially forming a work to be inspected. The work data includes, for example, the shape, dimensions, and surface state of the work. The work data is input as three-dimensional data, for example, by 3D-CAD data of the work. The work data acquisition unit 21 may acquire not only the work data but also peripheral data. Peripheral data is data serving as a basis for forming an inspection condition in a pseudo manner. The peripheral data includes data relating to the inspection environment that changes every moment in the inspection facility, such as the direction, intensity, and color temperature of the light that irradiates the workpiece.

  The pseudo 3D image generation unit 22 generates a pseudo 3D image of the work from the work data acquired by the work data acquisition unit 21. Thereby, the pseudo 3D image generation unit 22 generates a pseudo 3D image that approximates a workpiece to be inspected by actual inspection equipment. In this case, the pseudo 3D image generation unit 22 may generate a pseudo 3D image in consideration of peripheral data. As described above, the pseudo 3D image generation unit 22 generates a simulated work as a pseudo 3D image based on the work data and, if necessary, the peripheral data.

  The 2D defective partial image generation unit 23 generates a 2D defective partial image that virtually represents a defect occurring in the workpiece. The 2D defective partial image is a two-dimensional image that represents scratches, discoloration, and the like that are considered as defects occurring in the workpiece. The 2D defective partial image generation unit 23 can create an arbitrary 2D defective partial image by an input from the input unit 28, for example. The 2D defective partial image generation unit 23 accumulates the created 2D defective partial image in, for example, a storage unit (not illustrated) of the control device 11. Therefore, the 2D defective partial image generation unit 23 can also acquire an arbitrary 2D defective partial image by reading the 2D defective partial image accumulated in the storage unit.

  The negative case image combining unit 24 pastes the 2D defective partial image generated by the 2D defective partial image generating unit 23 on the pseudo 3D image generated by the pseudo 3D image generating unit 22. Thereby, the negative case image composition unit 24 synthesizes a 3D negative case image of a pseudo work in which a 2D defective partial image is pasted on a part of the pseudo 3D image. That is, the negative case image composition unit 24 seems to have attached a sticker of a 2D defective part image representing a two-dimensional defective part to a part of a three-dimensional image of a pseudo work generated as a pseudo 3D image. 3D negative case images are synthesized.

  The 2D camera image creation unit 25 creates a pseudo 2D image from the 3D negative case image generated by the negative case image synthesis unit 24 by synthesizing the pseudo 3D image and the 2D defective partial image. Specifically, the 2D camera image creation unit 25 shoots the 3D negative case image generated by the negative case image synthesis unit 24 in a pseudo manner from a specified range that matches the shooting range of the camera 102 of the actual visual inspection device 100. To do. In this case, the 2D camera image creation unit 25 does not photograph a 3D negative case image with an actual camera, but simulates a 3D negative case image created in a pseudo manner from a designated range that matches the photographing range of the camera 102. Take a picture. As a result, the 2D camera image creation unit 25 creates a pseudo 2D image in which a 3D negative case image is captured in a pseudo manner by software. Generation of a pseudo 3D image by the pseudo 3D image generation unit 22, creation of a 2D defective partial image by the 2D defective partial image generation unit 23, synthesis of a 3D negative case image by the negative case image synthesis unit 24, and a 2D camera image creation unit The creation of the pseudo 2D image by 25 is executed as a series of processes in terms of software.

  The 2D image transmission unit 26 transmits the pseudo 2D image created by the 2D camera image creation unit 25 to the inspection apparatus 13 as pseudo 2D image data. The stored data transmission unit 27 transmits the pseudo 2D image data created by the 2D camera image creation unit 25 and the creation condition data of the pseudo 2D image to the storage device 14. The creation condition data includes data relating to conditions set for creating a pseudo 2D image, such as work data used to create a pseudo 2D image by the 2D camera image creation unit 25 and data indicating the type of 2D defective partial image. Contains. As a result, when the pseudo 2D image is created by the 2D camera image creation unit 25, the 2D image transmission unit 26 transmits the pseudo 2D image data of the pseudo 2D image to the inspection apparatus 13, and the stored data transmission unit 27 performs the pseudo 2D image data. The 2D image data is transmitted to the storage device 14 together with the creation condition data.

  The inspection device 13 corresponds to the inspection means in the claims, and includes an inspection program acquisition unit 31, a program execution unit 32, a 2D image assignment unit 33, and an inspection result transmission unit. The inspection program acquisition unit 31 acquires an inspection program stored in a storage unit (not shown) of the control device 11, for example. The inspection program is set in advance for inspection of the workpiece 104 in the actual visual inspection apparatus 100. Therefore, the inspection program acquisition unit 31 acquires an inspection program that is set in advance and stored in a storage unit (not shown). The inspection program may be set as one of subroutines included in the evaluation program. In this case, the inspection program acquisition unit 31 acquires an inspection program from the evaluation program.

  The program execution unit 32 executes the inspection program acquired by the inspection program acquisition unit 31. The 2D image adding unit 33 gives a pseudo 2D image as an inspection target pseudo 2D image to the inspection program executed by the program execution unit 32. That is, the 2D image providing unit 33 provides the pseudo 2D image data transmitted from the 2D image transmission unit 26 of the image generation device 12 to the inspection program executed by the program execution unit 32 as inspection target pseudo 2D image data. . The program execution unit 32 inspects the inspection target pseudo 2D image based on the inspection target pseudo 2D image data supplied from the 2D image adding unit 33 by executing the inspection program. Then, the inspection result transmission unit 34 transmits the inspection result obtained as a result of executing the inspection program to the storage device 14. That is, when the inspection target pseudo 2D image is inspected by executing the inspection program, the inspection result transmission unit 34 transmits the inspection result to the storage device 14 as inspection result data.

  The storage device 14 includes a storage area unit 41, a receiving unit 42 and a corresponding unit 43. The storage area unit 41 is configured by a large-capacity storage unit such as an HDD, and stores various data transmitted from the image generation device 12 and the inspection device 13. The receiving unit 42 acquires various types of data from the storage data transmission unit 27 of the image generation device 12 and the inspection result transmission unit 34 of the inspection device 13 and writes the acquired various types of data in the storage area unit 41. Specifically, the receiving unit 42 acquires pseudo 2D image data and creation condition data from the stored data transmission unit 27 of the image generation device 12. Further, the receiving unit 42 acquires inspection result data from the inspection result transmitting unit 34 of the inspection device 13.

  The correspondence unit 43 associates, that is, associates various data acquired by the reception unit 42. Specifically, the correspondence unit 43 adds the pseudo 2D image data and the creation condition data transmitted from the storage data transmission unit 27 of the image generation device 12 to the pseudo 2D image transmitted from the inspection result transmission unit 34 of the inspection device 13. The inspection result data corresponding to the image, that is, the inspection target pseudo 2D image is associated and stored in the storage area unit 41. That is, as shown in FIG. 2, the pseudo 2D image data 52 of the inspection target pseudo 2D image 51 that is the inspection target in the inspection apparatus 13, the creation condition data 53 of the inspection target pseudo 2D image 51, and the inspection target pseudo 2D The image inspection result data 54 is stored in the storage area unit 41 as one inspection data file 55.

  The display device 15 is composed of, for example, a liquid crystal display. The display device 15 displays images and characters based on the inspection data file 55 stored in the storage device 14. Specifically, the display device 15 displays an image based on the pseudo 2D image data 52 constituting the inspection data file 55 shown in FIG. 2, a creation condition based on the creation condition data 53, and an inspection result based on the inspection result data 54. To do.

Next, the flow of processing of the evaluation system 10 having the above configuration will be described with reference to FIGS.
When the evaluation of the visual inspection apparatus is started by the evaluation system 10, the work data acquisition unit 21 of the image generation apparatus 12 acquires work data via the input unit 28 (S101). At this time, the work data acquisition unit 21 may acquire peripheral data together. In the workpiece data, the shape of the workpiece is expressed in three dimensions.

  When the work data is acquired by the work data acquisition unit 21, the pseudo 3D image generation unit 22 of the image generation device 12 generates a pseudo 3D image of the work from the acquired work data (S102). The 3D image generation unit 22 virtually constructs a visual inspection apparatus 100 as shown in FIG. 10, for example, and executes a simulation based on the work data by the virtual visual inspection apparatus 100, thereby performing pseudo 3D of the workpiece. Generate an image. Specifically, the 3D image generation unit 22 creates a pseudo 3D image 62 of the work from the work data 61 acquired by the work data acquisition unit 21. In the pseudo 3D image 62, the workpiece 104 to be inspected by the actual visual inspection apparatus 100 is faithfully reproduced based on the workpiece data 61. Note that the pseudo 3D image 62 and the visual inspection device 100 are generated by virtual simulation to the last, and are not necessarily generated visually.

  When the pseudo 3D image 62 is generated, the negative case image synthesis unit 24 synthesizes the 3D negative case image 63 (S103). The negative case image combining unit 24 pastes the 2D defective partial image 64 generated by the 2D defective partial image generating unit 23 on the pseudo 3D image 62 generated by the 3D image generating unit 22. As a result, the negative case image composition unit 24 synthesizes the 3D negative case image 63 of the pseudo work in which the 2D defective partial image 64 is pasted on a part of the pseudo 3D image 62. As shown in FIG. 5, the 2D defective portion image 64 includes a colored defective portion 65 that expresses scratches, discoloration, and the like, and a colorless and transparent background 66 that is set in a predetermined range surrounding the defective portion 65. Yes. The background 66 is set to an arbitrary shape including the defective portion 65 such as a polygonal shape such as a quadrangle, or a circle or an ellipse. The 2D defective partial image generation unit 23 generates the 2D defective partial image 64. The 2D defective portion image 64 is a 2D planar image, but is shaded in consideration of, for example, light emitted from the light source by changing the color and tone of the defective portion 65. Therefore, the 2D defective portion image 64 three-dimensionally represents the defective portion 65 on the 2D plane.

  When the 3D negative case image 63 is generated, the 2D camera image creation unit 25 creates a pseudo 2D image 67 based on the generated 3D negative case image 63 (S104). For example, as illustrated in FIG. 10, the 2D camera image creation unit 25 simulates an image obtained by virtually capturing the 3D negative case image 63 reproduced by the virtual negative case image synthesis unit 24 from a predetermined camera position. A pseudo 2D image 67 is created. That is, the 2D camera image creation unit 25 virtually shoots the 3D negative case image 63 from the same position where the camera 104 of the actual visual inspection apparatus 100 shown in FIG. Then, the 2D camera image creation unit 25 creates a virtually captured image as a pseudo 2D image 67. This pseudo 2D image 67 is also created by simulation based on the 3D negative case image 63 of a virtual work, and is not actually photographed using the camera 102. The pseudo 2D image 67 is created as a visible image by simulation based on the 3D negative case image 63.

  On the other hand, the inspection apparatus 13 acquires an inspection program as the evaluation system 10 starts evaluation of the visual inspection apparatus 100 (S201). The inspection program acquisition unit 31 acquires a predetermined inspection program stored in a storage unit (not shown). Then, the program execution unit 32 activates the inspection program acquired by the inspection program acquisition unit 31 (S202). When the inspection program is executed by the program execution unit 32, the 2D image adding unit 33 determines whether or not the pseudo 2D image 67 is transmitted from the image generation device 12 (S203). When there is no transmission of the pseudo 2D image 67 from the image generation device 12 (S203: No), the 2D image adding unit 33 waits until the pseudo 2D image 67 is transmitted.

  The storage device 14 is activated when the evaluation system 10 starts the evaluation of the visual inspection device 100 (S301). Then, the receiving unit 42 of the storage device 14 determines whether or not the pseudo 2D image data of the pseudo 2D image 67 is transmitted from the image generation device 12 (S302). When there is no transmission of these data from the image generation device 12 (S302: No), the reception unit 42 stands by until there is transmission of these data.

  When the pseudo 2D image 67 is created by the 2D camera image creation unit 25 in S104, the 2D image transmission unit 26 of the image generation device 12 creates the created pseudo 2D image 67, that is, pseudo 2D image data that is data of the pseudo 2D image 67. Is transmitted to the inspection apparatus 13 (S105). Further, the stored data transmission unit 27 of the image generation device 12 transmits the created pseudo 2D image data of the pseudo 2D image 67 to the storage device 14 (S106). When the pseudo 2D image data is transmitted from the 2D image transmission unit 26 to the inspection device 13 in S105, the 2D image adding unit 33 of the inspection device 13 acquires the pseudo 2D image data (S204). In addition, when the pseudo 2D image data is transmitted from the storage data transmission unit 27 to the storage device 14 in S105, the reception unit 42 of the storage device 14 acquires the pseudo 2D image data (S303). Then, the receiving unit 42 of the storage device 14 determines whether or not the inspection result data 54 of the pseudo 2D image 67 is transmitted from the inspection device 13 (S304). When there is no transmission of the inspection result data 54 from the inspection device 13 (S304: No), the reception unit 42 waits until the inspection result data 54 is transmitted.

  Upon acquiring the pseudo 2D image data of the pseudo 2D image 67 from the 2D image transmission unit 26 of the image generation device 12 in S204, the 2D image adding unit 33 of the inspection device 13 executes the acquired pseudo 2D image data by the program execution unit 32. The inspection target pseudo 2D image data is given to the inspection program (S205). Thereby, the program execution unit 32 executes the inspection of the pseudo 2D image based on the given inspection target pseudo 2D image data, that is, the inspection target pseudo 2D image 51 based on the predetermined inspection program (S206). The program execution unit 32 performs inspection of the inspection target pseudo 2D image 51 by comparing the inspection target pseudo 2D image 51 with a preset master image. The master image is an image serving as a reference for inspection, that is, an image of a positive case, and is created in advance under predetermined conditions. When executing the inspection in S206, the program execution unit 32 creates inspection result data 54 as the inspection result (S207). The inspection result data 54 includes, for example, a determination result indicating whether the inspection target pseudo 2D image 51 of the workpiece subjected to the inspection is a non-defective product that has passed the inspection or a defective product that has failed the inspection. Yes.

  When the inspection result data 54 is created by the program execution unit 32 in S207, the inspection result transmission unit 34 of the inspection device 13 transmits the created inspection result data 54 to the storage device 14 (S208). When the inspection result data 54 is transmitted from the inspection result transmitting unit 34 to the storage device 14 in S208, the receiving unit 42 of the storage device 14 acquires the inspection result data 54 (S305). When the storage unit 14 acquires the inspection result data 54 from the inspection device 13 in the receiving unit 42, the storage unit 14 obtains the pseudo 2D image data 52 acquired from the image generation device 12 in S 303 and the inspection result data 54 acquired in S 305 in the corresponding unit 43. Correspondingly, the data is stored in the storage area 41 as the inspection data file 55 (S306). The correspondence unit 43 writes the pseudo 2D image data 52 acquired in S303 and the inspection result data 54 acquired in S305 to the storage area unit 41 as one inspection data file 55. After transmitting the pseudo 2D image data 52 to the storage device 14 in S106, the image generation device 12 determines whether or not the next inspection target is set (S107). When the next inspection target is set (S107: Yes), the image generating apparatus 12 returns to S101 and repeats the processing after S101. When the next inspection target is not set (S107: No), the image generation device 12 ends the process.

  When the evaluation of the visual inspection device 100 by the evaluation system 10 is started, the display device 15 displays necessary information. Then, based on the inspection data file 55 stored in S306, images and characters are displayed as an evaluation screen. That is, the display device 15 displays images and characters based on the pseudo 2D image data 52 and the inspection result data 54 as an evaluation screen. Further, the display device 15 displays the pseudo 2D image data 52 and the inspection result data 54 every time the work data is changed in S101.

(Image accuracy)
Next, the accuracy of the pseudo 2D image 67 based on the 3D negative case image 63 of the work generated by the above-described image generation apparatus 12 with respect to the actual image of the work 104 actually captured using the camera 102 will be described.
In order to verify the accuracy of the pseudo 2D image 67, an image of a real work with physical scratches was compared with a pseudo 2D image 67 in which the 2D defective partial image 64 was pasted on the pseudo 3D image 62. Specifically, a workpiece with a scratch is obtained by physically scratching the actual workpiece of the positive case. FIG. 6A shows an actual positive case image 201 obtained by photographing a positive case work, and FIG. 6B shows an actual negative case image 202 obtained by photographing a work in which the positive case shown in FIG. It is. 6C is a pseudo-positive case image 203 showing a pseudo 3D image 62 formed from work data, and FIG. 6D is a defective partial image 204 showing a 2D defective partial image 64. The pseudo 3D image 62 shown in FIG. 6C is pasted with the 2D defective partial image 64 shown in FIG. 6D to synthesize the image, and then simulated by the camera from the same position as in FIG. 6B. The obtained image becomes a pseudo negative case image 205. This pseudo negative case image 205 corresponds to the pseudo 2D image 67.

  For verification, the difference between the real positive case image 201 shown in FIG. 6A and the real negative case image 202 shown in FIG. 6B is expressed as the pseudo positive case image 203 shown in FIG. It compared with the difference with the pseudo negative example image 205 shown to (E). The difference comparison procedure is as follows. That is, as shown in FIG. 7, each of the positive case image 301 and the negative case image 302 is binarized at an arbitrary threshold value X set in advance. Then, a difference is extracted from the binary positive case image 303 and the binary negative case image 304 obtained by binarization, and a binary difference image 305 is generated.

  For example, as shown in FIG. 8, in the case of an image obtained by photographing a real work, the actual positive case image 201 and the real negative case image 202 are binarized using a predetermined threshold value X, respectively. As a result, a real binary positive case image 201a and a real binary negative case image 202a are obtained. A real binary difference image 400 is obtained from the difference between the real binary positive case image 201a and the real binary negative case image 202a. Similarly, for example, as shown in FIG. 9, in the case of a pseudo image generated by the image generation device 12, the pseudo positive case image 203 and the pseudo negative case image 205 are binarized using the same threshold value X as the real thing. The Thereby, the pseudo binary positive case image 203a and the pseudo binary negative case image 205a are obtained. A pseudo binary difference image 500 is obtained from the difference between the pseudo binary positive case image 203a and the pseudo binary negative case image 205a.

  When the actual binary difference image 400 obtained by the above processing is compared with the pseudo binary difference image 500, it is clear from the actual binary difference image 400 shown in FIG. 8 and the pseudo binary difference image 500 shown in FIG. As described above, the defective portion, that is, the portion extracted as a scratch substantially matches. As a result, the 2D defective partial image 64 is pasted on the pseudo 3D image 62 to synthesize the 3D negative case image 63, and the pseudo 3D negative image 63 (pseudo positive case image 203) It was verified that the same inspection result as that of the negative case work generated by physically scratching the positive case work can be obtained.

According to the embodiment described above, the negative case image synthesizing unit 24 adds the 2D defective partial image generated by the 2D defective partial image generating unit 23 to the pseudo 3D image 62 of the workpiece generated pseudo by the 3D image generating unit 22. 64 is pasted. The evaluation of the visual inspection device 100 is finally performed using an image obtained by photographing the workpiece 104, that is, a 2D image even if the actual workpiece 104 is used. For this reason, even if the 2D defective partial image 64 is pasted on the pseudo 3D image 62 of the workpiece, the resulting captured image is a pseudo 2D image 67 that is substantially the same as the image of the actual workpiece 104 captured. Further, the 2D defective part image 64 representing the defective part can be arbitrarily expressed in consideration of the imaging direction of the workpiece in the visual inspection device 100, for example. When the pseudo 2D image 67 is used for the evaluation of the visual inspection device 100, the image actually captured by the camera 102 is also a 2D image without a depth. As a result, even if the 2D defective part image 64 is pasted on the pseudo 3D image 62 of the pseudo work, a pseudo 2D image 67 equivalent to that obtained by photographing the work having the 3D defective part is created. As a result, by accumulating many defective parts such as scratches and discoloration as the 2D defective part image 64, it is possible to easily create a negative case by simply attaching the 2D defective part image 64 to the pseudo 3D image 62 of the pseudo work. A 3D negative case image 63 corresponding to the work is created. By using the 3D negative case image 63, the visual inspection apparatus 100 is evaluated for a wide range of negative case works. Therefore, it is possible to evaluate a wide range of negative case works in the visual inspection device 100 without obtaining real works of positive cases and negative cases, and the inspection accuracy of the visual inspection device 100 can be improved.
The present invention described above is not limited to the above-described embodiment, and can be applied to various embodiments without departing from the gist thereof.

  In the drawings, 10 is an evaluation system, 12 is an image generation device, 13 is an inspection device (inspection means), 14 is a storage device, 15 is a display device, 21 is a work data acquisition unit, 22 is a 3D image generation unit, and 23 is 2D. A defective partial image generation unit, 24 is a negative case image synthesis unit, and 25 is a 2D camera image creation unit.

Claims (1)

An evaluation system that evaluates a visual inspection device that inspects a workpiece based on a 2D captured image captured by the camera by capturing a workpiece to be inspected with a camera provided in an inspection facility,
Work data acquisition means for acquiring work data serving as a basis for artificially forming the work;
3D image generation means for generating a pseudo 3D image of the work from the work data acquired by the work data acquisition means;
2D defective part image generation means for generating a 2D defective part image that includes a colored defective part and a colorless and transparent background of a predetermined shape and that virtually expresses a defect occurring in the workpiece;
The 2D defective partial image generated by the 2D defective partial image generating unit is pasted on the pseudo 3D image generated by the 3D image generating unit, and a pseudo 3D negative case image of the workpiece having the defective part is synthesized. Negative case image synthesizing means,
2D camera image creation means for creating a pseudo 2D image in which the 3D negative case image to which the 2D defective partial image is pasted is photographed in a pseudo manner from a designated range that matches the photographing range of the camera;
The pseudo 2D image created by the 2D camera image creating means is acquired as an inspection target pseudo 2D image, and inspection means for inspecting the workpiece from the inspection target pseudo 2D image according to a preset inspection program;
An evaluation system for a visual inspection apparatus, comprising: