Quality control program, quality control equipment, and quality control method

Abstract

While conventional methods can identify the location of foreign objects on the inspected item, they are prone to false detections. [Solution] The quality control program of the present invention takes an input image set containing multiple images of items to be inspected as input, and when the items to be inspected are determined to be good, the feature quantity extraction process outputs feature quantity data that includes detection features where the difference from the reference feature is larger for parts of the items to be inspected that differ in shape and color from the items to be inspected that are determined to be good, and causes the computer to execute the following: A coordinate transformation mask generated from the shooting position and shooting angle when the images to be inspected were taken is applied to the feature quantity data, and a three-dimensional alignment process is performed to generate a three-dimensional feature map by integrating multiple detection features in the multiple feature data generated corresponding to the multiple images to be inspected in the input image set that have the same three-dimensional position as the items to be inspected.

Description

Technical Field

[0001] The present invention relates to a quality control device, a quality control program, and a quality control method, and more particularly, to a quality control device, a quality control program, and a quality control method for detecting defects of an inspection target based on a product image obtained from the inspection target product.

Background Art

[0002] In a manufacturing process for producing products, defect inspections are performed to check for the presence or absence of defects in the produced products at various times. Such defect inspections are often performed using image analysis. Therefore, an example of such a defect inspection using image analysis is disclosed in Patent Document 1.

[0003] The foreign object detection device described in Patent Document 1 includes an electromagnetic wave generation unit that emits electromagnetic waves from a point light source, a rotary holding unit that holds an inspection target and rotates it around a rotation axis, a detection unit that detects the electromagnetic waves transmitted through the inspection target, and a trajectory extraction unit that extracts the trajectory of a foreign object in the inspection target based on the rotation image data of the inspection target generated by detecting the electromagnetic waves transmitted through the inspection target while the inspection target is rotated around the rotation axis, and calculates the coordinates of the center (C2) of the circle including the trajectory. The device further includes a foreign object position specifying unit that specifies the position of the foreign object in the direction of the rotation axis based on the coordinates of the center (C2).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In the foreign object detection device described in Patent Document 1, the position of the detected foreign object is determined by considering the amount of rotation, and the location of the foreign object on the inspected item is confirmed using detection obtained by observing the inspected item from multiple directions. In this way, when the inspected item is observed from multiple directions, the position of the captured foreign object can be determined by considering the angle of capture and other factors, thereby determining the location of the foreign object on the inspected item. However, while the foreign object detection device described in Patent Document 1 can identify the location of foreign objects on the product being inspected, there is a problem in that it may cause false detections in areas where the appearance changes depending on the angle. [Means for solving the problem]

[0006] One aspect of the quality control program described herein involves having a computer perform the following: an input image set including multiple images of an item to be inspected obtained by photographing a single item to be inspected from multiple different angles; a feature extraction process that outputs feature data including detection features where the difference from the reference feature is larger for parts of the item that differ in shape and color from the item judged to be good; and a three-dimensional alignment process that applies a coordinate transformation mask generated from the shooting position and shooting angle when the images of the item to be inspected were taken to the feature data, and integrates multiple detection features in the feature data generated in response to the multiple images of the item to be inspected in the input image set where the three-dimensional position of the item to be inspected is the same, thereby generating a three-dimensional feature map.

[0007] One embodiment of the quality control device according to this disclosure includes: a feature extraction unit that takes as input an input image set including multiple images of an item to be inspected obtained by photographing one item to be inspected from multiple different angles, and outputs feature data including detection features in which the difference from the reference feature is larger the more the shape and color differ from the item to be inspected that is judged to be a good product; and a three-dimensional alignment processing unit that applies a coordinate transformation mask generated from the shooting position and shooting angle when the images to be inspected were taken to the feature data, and integrates multiple detection features in the multiple feature data generated corresponding to the multiple images to be inspected in the input image set in which the three-dimensional position of the item to be inspected is the same, thereby generating a three-dimensional feature map.

[0008] One aspect of the quality control method relating to this disclosure is a feature extraction process that takes an input image set containing multiple images of an item to be inspected obtained by photographing one item to be inspected from multiple different angles as input, and outputs feature data containing detection features where the difference from the reference feature is larger in areas where the shape and color differ from the item to be inspected that is judged to be good, when the feature output when the item to be inspected is judged to be good is used as a reference feature; and a three-dimensional alignment process that applies a coordinate transformation mask generated from the shooting position and shooting angle when the images to be inspected were taken to the feature data, and integrates multiple detection features in the multiple feature data generated corresponding to the multiple images to be inspected in the input image set where the three-dimensional position of the item to be inspected is the same, thereby generating a three-dimensional feature map, all of which are performed by computer automation.

[0009] The quality control program, quality control device, and quality control method disclosed herein perform alignment processing to convert defective areas captured in multiple images into positions on the product under inspection, and then integrate the feature quantities of the defective areas after alignment into a single value to generate a three-dimensional feature map that includes a single detected feature quantity that includes the characteristics of changes in detected feature quantities due to the shooting conditions. [Effects of the Invention]

[0010] According to the quality control program, quality control device, and quality control method disclosed herein, it becomes possible to present detection results that include information about defects where the appearance changes depending on the viewing angle. [Brief explanation of the drawing]

[0011] [Figure 1] This is a schematic diagram of the quality control system according to Embodiment 1. [Figure 2] This figure illustrates an example of an input image set generated by the quality control system according to Embodiment 1. [Figure 3] This is a block diagram of the quality control device according to Embodiment 1. [Figure 4] This is a block diagram of the quality control device according to Embodiment 2. [Figure 5] This is a block diagram of the quality control device according to Embodiment 3. [Figure 6] This is a block diagram of the quality control device according to Embodiment 4. [Figure 7] This is a block diagram of the quality control device according to Embodiment 5. [Modes for carrying out the invention]

[0012] For clarity of explanation, the following descriptions and drawings have been omitted and simplified as appropriate. Furthermore, each element shown in the drawings as a functional block performing various processes can be composed of a CPU (Central Processing Unit), memory, and other circuits in hardware terms, and implemented in software terms by programs loaded into memory. Therefore, it will be understood by those skilled in the art that these functional blocks can be implemented in various ways using hardware alone, software alone, or a combination thereof, and are not limited to any one of these. In each drawing, the same elements are denoted by the same reference numeral, and redundant explanations have been omitted where necessary.

[0013] Furthermore, the program described above includes, when loaded into a computer, a set of instructions (or software code) for causing the computer to perform one or more of the functions described in the embodiments. The program may be stored in a non-temporary computer-readable medium or a physical storage medium. Examples, but not limited to, include random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technologies, CD-ROM, digital versatile disc (DVD), Blu-ray® disc or other optical disc storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices. The program may be transmitted over a temporary computer-readable medium or a communication medium. Examples, but not limited to, include temporary computer-readable medium or a communication medium that includes electrically, optically, acoustically, or otherwise propagating signals.

[0014] Embodiment 1 The quality control system 1 according to Embodiment 1 aims to detect defects in products under inspection using image analysis technology. Therefore, the quality control system 1, which includes a configuration for photographing products under inspection and a quality control device that determines whether the products are good or bad based on the captured images, will be described. Figure 1 shows a schematic diagram of the quality control system 1 according to Embodiment 1.

[0015] In the example shown in FIG. 1, the quality management system 1 according to Embodiment 1 includes a quality management device 10, a camera 11, a lighting device 12, and a turntable 13. The quality management system 1 according to Embodiment 1 places the inspection target product OBJ on the turntable 13 and rotates the inspection target product OBJ by a predetermined rotation angle, and photographs the inspection target product OBJ with the camera 11 for each rotation angle, thereby obtaining a plurality of inspection target images of the inspection target product OBJ photographed from various angles. Also, in the example shown in FIG. 1, the lighting device 12 is shown. The lighting device 12 illuminates the inspection target product OBJ so that the image of the inspection target product OBJ is more clearly reflected in the inspection target image.

[0016] Note that, in the example shown in FIG. 1, an example is shown in which the inspection target product OBJ is rotated to photograph the inspection target product OBJ at a number of shooting angles to obtain a plurality of inspection target images. However, without rotating the inspection target product OBJ, a plurality of cameras 11 and lighting devices 12 may be arranged around the inspection target product OBJ to obtain a plurality of inspection target images from a number of cameras 11. Also, it is preferable that the imaging system is configured so that the inspection target product OBJ can be photographed from more angles, such as the upper surface, lower surface, diagonal direction, etc., rather than from the side surface of the inspection target product OBJ.

[0017] In the quality management system 1 according to Embodiment 1, a plurality of inspection target images obtained by photographing one inspection target product from a plurality of shooting angles are used as one input image set. Then, the quality management device 10 outputs one three-dimensional feature amount data for one input image set. This three-dimensional feature amount data is, for example, data labeled so that the value of the portion corresponding to the defective area, which has a different shape and color from the non-defective product, has a large difference from the reference feature amount that is the feature amount of the non-defective product, and the three-dimensional position of the inspection target product can be known. More specifically, the three-dimensional feature amount data is a three-dimensional feature amount map in which each of the detection feature amounts is mapped to the three-dimensional position of the inspection target product based on the label attached to the detection feature amount. Hereinafter, the configuration and generation method for generating this three-dimensional feature amount map will be described.

[0018] First, the input image set will be described. FIG. 2 shows a diagram for explaining an example of the input image set generated by the quality management system according to Embodiment 1. The example of the input image set shown in FIG. 2 includes eight inspection target images taken by changing the shooting angle by 45° each around the rotation axis extending in the longitudinal direction of the inspection target product OBJ with respect to the inspection target product OBJ. Note that the example shown in FIG. 2 is an example of the input image set, and the number of images included in the input image set is not limited to eight, and will be nine or more if the number of shooting angles increases.

[0019] Also, in the example shown in FIG. 2, defective regions A to E are shown. The defective region A is, for example, a stain, and there is little change in color regardless of the change in the shooting angle. The defective region B is, for example, a dent, and the luminance (color) changes greatly as the shooting angle changes. The defective region C is, for example, fiber dust or processing scraps, and the shape changes as the shooting angle changes. The defective region D is, for example, a defect on a protrusion such as granular dust, molding defect, processing defect, burr, etc., and the shape changes as the shooting angle changes. The defective region E is, for example, a concave defect such as a cavity generated during casting, and the shape changes as the shooting angle changes.

[0020] As described above, for the defects of the inspection target product, there are those in which the shape and color change by changing the shooting angle. Then, in the quality management device 10, a detection feature amount in which the difference from the reference feature amount becomes large due to the characteristics of the defective region is generated for each pixel of the inspection target image, and among the detection feature amounts generated based on the plurality of inspection target images included in one input image set, a plurality of detection feature amounts having the same three-dimensional position of the inspection target product are integrated using statistical processing to generate three-dimensional feature amount data. That is, by verifying the magnitude of the detection feature amount included in the three-dimensional feature amount data, the user can determine the position of the defect, the size of the defective region, and the type of the defect.

[0021] Next, the quality control device 10 will be described in detail. The quality control device 10 can be configured with dedicated hardware, or it can be implemented by running a quality control program on a computer equipped with a program execution unit. Below, we will describe an example of how the quality control program realizes the functions and processes described in the various functional blocks.

[0022] Figure 3 shows a block diagram of the quality control device according to Embodiment 1. Figure 3 shows the input image set input to the quality control device 10 and the three-dimensional feature data output by the quality control device 10. As shown in Figure 3, the quality control device 10 has a defect detection unit 20 and a three-dimensional alignment processing unit 22. The defect detection unit 20 has a feature extraction unit 21. The input image set includes multiple inspection target images obtained by photographing a single inspection target from multiple different angles. The input image set also includes an XYZ mask that includes at least the shooting angle information at the time the inspection target image was taken.

[0023] The feature extraction unit 21 takes the input image set as input and outputs feature data that includes detection features, where the difference from the reference feature is larger for areas where the shape and color differ from those of an item judged as good, using the feature output when the item to be inspected is judged as good as a reference feature. In one example, these detection features are generated on a pixel-by-pixel basis in an image with the same number of pixels as the image to be inspected. In other words, the feature data output by the feature extraction unit 21 is a mapping of the detection features onto a two-dimensional image. To put it another way, the feature data output by the feature extraction unit 21 is a mapping of detection features with a large difference from the reference feature to areas in the item OBJ captured in the input image to be inspected that have a different shape or color from a good item.

[0024] Furthermore, the feature extraction unit 21 can use a classifier that outputs detection features that abstract the input values ​​based on a pre-generated transfer function, or it can use a classifier to which trained parameters generated by machine learning are applied. An example of a classifier to which trained parameters are applied is one that outputs the similarity (distance) between the input values ​​and all the training data as detection features using an algorithm based on the k-nearest neighbors method.

[0025] The three-dimensional alignment processing unit 22 applies a coordinate transformation mask (e.g., an XYZ mask) generated from the shooting position and shooting angle when the inspection target image was captured to the feature data output by the feature extraction unit 21. It then integrates multiple detection features that correspond to the same three-dimensional position of the inspection target among the multiple feature data generated for multiple inspection target images included in the input image set into a single value to generate a three-dimensional feature map (e.g., three-dimensional feature data). In other words, the three-dimensional feature data generated using the three-dimensional alignment processing unit 22 is a single detection feature that integrates multiple detection features and maps them to each position of the inspection target.

[0026] Here, the synthesis process, which combines multiple detection features with the same three-dimensional position as the object being inspected into a single value, uses statistical methods to combine multiple values ​​into one value while maintaining as much of the meaning of each value as possible. Examples of synthesis processes include averaging, which takes the average of multiple values, and function processing, which combines multiple values ​​into a single value using a predetermined function.

[0027] As described above, by using the quality control device 10 according to Embodiment 1, feature data is generated from each of the multiple images of an inspected product taken by changing the shooting conditions such as the shooting angle and lighting angle for a single inspected product. This feature data includes detection features that are characteristic values ​​of the parts that differ from good products. Three-dimensional feature data is generated by integrating multiple detection features from the multiple feature data that have the same three-dimensional position as the inspected product into a single value. In other words, the quality control device 10 generates three-dimensional feature data that includes detection features capable of determining the location, size, and type of defect. This makes it possible to determine the location, size, and type of defect by referring to the three-dimensional feature data.

[0028] Defects can range from dust that can be removed after manufacturing to dents, scratches, and voids that are difficult to remove even in later processes, as well as shape defects that are unacceptable according to the product specifications. Therefore, by enabling quality control that can identify the type of defect, it becomes possible to improve production efficiency by preventing products that should not be considered defective from being classified as such, such as by correcting minor defects in later processes and removing products with serious defects.

[0029] Embodiment 2 Embodiment 2 describes a quality control device 10a, which is a modified version of the quality control device 10. In the description of Embodiment 2, components that are the same as those described in Embodiment 1 are denoted by the same reference numerals as in Embodiment 1 and their descriptions are omitted.

[0030] Figure 4 shows a block diagram of the quality control device 10a according to Embodiment 2. As shown in Figure 4, the quality control device 10a is a quality control device 10 with the addition of a defect determination unit 23. The defect determination unit 23 presents to the user the result of determining whether a product under inspection is defective and the type of defect, based on at least one of the shape, size, and size of the detected feature in the defective region, where the detected feature that makes a large difference from the reference feature is mapped in a three-dimensional feature map (e.g., three-dimensional feature data). For example, the defect determination unit 23 identifies the type of defect by providing the shape, size, and detected feature of the defective region to a predetermined function and a table that associates the shape and size of the defective region with the size of the detected feature in the defective region.

[0031] As described above, in Embodiment 2, by using the defect determination unit 23, it is possible to obtain a determination result for the presence or absence of defects and the type of defects that does not depend on human judgment. In this way, by determining the presence or absence of defects and the type of defects through computer processing based on pre-set rules, fluctuations in the judgment criteria can be eliminated. Furthermore, by performing computer processing, it is possible to increase the number of inspected items that can be judged as good or bad within a certain period of time, thereby improving productivity.

[0032] Embodiment 3 Embodiment 3 describes a quality control device 10b, which is a modified version of the quality control device 10a. In the description of Embodiment 3, components that are the same as those described in Embodiments 1 and 2 are denoted by the same reference numerals as in Embodiments 1 and 2, and their descriptions are omitted.

[0033] Figure 5 shows a block diagram of the quality control device 10b according to Embodiment 3. As shown in Figure 5, the quality control device 10b is a modified version of the quality control device 10a in which the defect detection unit 20 is replaced with a defect detection unit 30. The defect detection unit 30 is a modified version of the defect detection unit 20 with the addition of an image processing unit 31. The defect detection unit 30 performs image processing to output a detection result image as feature data, which maps the detected features contained in the feature data output by the feature extraction unit 21 to the corresponding positions in the image to be inspected.

[0034] Furthermore, in this image processing, if the detected feature quantity is less than a predetermined threshold, a binarization process may be performed in which the detected feature quantity is replaced with a first value, and if the detected feature quantity is equal to or greater than the threshold, the detected feature quantity is replaced with a second value.

[0035] By performing image processing in this manner, the feature data output by the defect detection unit 30 can be an image in which the detected features are mapped to the corresponding pixels of the image under inspection. As a result, in the quality control device 10b according to Embodiment 3, the three-dimensional alignment processing in the three-dimensional alignment processing unit 22 can be performed using image processing technology, making the processing easier. Furthermore, by performing binarization processing, it is possible to reduce the processing load on the calculation unit when performing image processing.

[0036] Embodiment 4 Embodiment 4 describes a quality control device 10c, which is a modified version of the quality control device 10a. In the description of Embodiment 4, components that are the same as those described in Embodiments 1 and 2 are denoted by the same reference numerals as in Embodiments 1 and 2, and their descriptions are omitted.

[0037] Figure 6 shows a block diagram of the quality control device 10c according to Embodiment 4. As shown in Figure 6, the quality control device 10c is a modified version of the quality control device 10a in which the defect determination unit 23 is replaced with a defect determination unit 40. The defect determination unit 40 includes a known defect determination unit 41, an unknown defect determination unit 42, and a pass / fail determination unit 43. The known defect determination unit 41 performs a known defect determination process to determine the presence and type of the aforementioned defect of a predetermined type. The unknown defect determination unit 42 performs an unknown defect determination process to determine the presence or absence of the aforementioned defect whose type cannot be determined. The pass / fail determination unit 43 outputs a determination result indicating that the inspected product is a good product if both the known defect determination process and the unknown defect determination process determine that there is no defect. On the other hand, if both the known defect determination process and the unknown defect determination process determine that there is a defect, the pass / fail determination result indicates that the product is defective, the type of defect, or information indicating that it is different from a known defect.

[0038] As described above, in Embodiment 4, by having an unknown defect determination unit 42, it becomes possible to determine whether an item under inspection has a new type of defect, as well as a known defect that has been set in advance.

[0039] Embodiment 5 Embodiment 5 describes a quality control device 10d, which is a modified version of the quality control device 10a. In the description of Embodiment 5, components that are the same as those described in Embodiments 1 and 2 are denoted by the same reference numerals as in Embodiments 1 and 2, and their descriptions are omitted.

[0040] Figure 7 shows a block diagram of the quality control device 10d according to Embodiment 4. As shown in Figure 7, the quality control device 10d is a quality control device 10a with the addition of a region feature integration processing unit 50. In addition, the defect determination unit 23 is replaced with a defect determination unit 51 in the quality control device 10d.

[0041] The region feature integration processing unit 50 applies a region transformation mask, which indicates which part of the product under inspection each coordinate corresponds to, to the feature data output by the image processing unit 31 for each image under inspection. It then integrates multiple detected features contained in the same part of the multiple feature data generated for multiple images under inspection included in the input image set to generate a region-unit feature map (e.g., region-unit feature data). The defect determination unit 51 refers to the three-dimensional feature data and the region-unit feature data to determine whether the product under inspection is defective and what type of defect it is.

[0042] In this way, by making a quality judgment based on region-unit feature data in which detection features are integrated for each part of the inspected product, and three-dimensional feature data in which detection features are mapped at the pixel level, the accuracy of defect detection can be improved.

[0043] It should be noted that the present invention is not limited to the embodiments described above, and can be modified as appropriate without departing from the spirit of the invention. [Explanation of Symbols]

[0044] 1. Quality Management System 10, 10a~10d Quality control device 11 Cameras 12 Lighting 13 Rotating Stands 20, 30 Defect detection unit 21 Feature Extraction Unit 22 Three-dimensional alignment processing unit 23, 40, 51 Defective judgment section 31 Image Processing Unit 41. Known defect detection unit 42 Unknown defect judgment section 43. Quality Determination Unit 50 Region Feature Integration Processing Unit OBJ Inspection Target Products

Claims

1. A feature extraction process takes an input image set containing multiple images of an item to be inspected, taken from multiple different angles, as input, and when the item to be inspected is determined to be good, the feature quantity output is set as the reference feature quantity, and the feature quantity output when the item to be inspected is determined to be good is set as the reference feature quantity, and the feature quantity data output includes detection feature quantities where the difference from the reference feature quantity is larger for parts of the item to be inspected that differ in shape and color from the item to be inspected that is determined to be good, and A three-dimensional alignment process is performed by applying a coordinate transformation mask generated from the shooting position and shooting angle when the inspection target image was captured to the feature data, integrating multiple detected features from the multiple feature data generated corresponding to the multiple inspection target images included in the input image set, where the three-dimensional position of the inspection target is the same, into a single value, thereby generating a three-dimensional feature map. A quality control program that causes a computer to execute a command.

2. The aforementioned quality control program further, A quality control program according to claim 1, which causes a computer to perform a defect determination process that presents to the user the result of determining whether the inspected product is defective and the type of defect, based on at least one of the shape and size of the defective region where the detected feature, which has a large difference from the reference feature, is mapped in the three-dimensional feature map, and the size of the detected feature in the defective region.

3. The aforementioned defect detection process is: A known defect determination process that determines the presence and type of the aforementioned defects of a predetermined type, An unknown defect determination process that determines whether or not there is a defect whose type cannot be identified, A quality determination process that outputs a result indicating that the inspected product is a good product if both the known defect determination process and the unknown defect determination process determine that there are no defects, A quality control program according to claim 2, which performs the following:

4. The system further includes a region feature integration process that applies a region transformation mask to the feature data, indicating which part of the product to be inspected each coordinate corresponds to, for each of the images to be inspected, and integrates multiple detected features included in the same part of the multiple feature data generated for each of the multiple images to be inspected included in the input image set, thereby generating a region-unit feature map. The quality control program according to claim 2, wherein the defect determination process determines whether the inspected product is defective and the type of defect by referring to the three-dimensional feature map and the region-unit feature map.

5. The quality control program according to claim 1, wherein the feature extraction process performs image processing to output a detection result image as feature data, in which the detected features are mapped to corresponding positions in the image to be inspected.

6. The quality control program according to claim 5, wherein the feature extraction process performs a binarization process in the image processing, in which if the detected feature is less than a preset threshold, the detected feature is replaced with a first value, and if the detected feature is equal to or greater than the threshold, the detected feature is replaced with a second value.

7. The quality control program according to claim 1, wherein in the three-dimensional alignment process, multiple detection features captured in multiple images of the items to be inspected, and where the three-dimensional position of the items to be inspected is the same, are combined into a single value by statistical processing.

8. A feature extraction unit takes an input image set containing multiple images of an item to be inspected, obtained by photographing a single item to be inspected from multiple different angles, as input, and when the feature quantity output when the item to be inspected is determined to be good is used as the reference feature quantity, it outputs feature quantity data that includes detection feature quantities where the difference from the reference feature quantity is larger for parts of the item to be inspected that differ in shape and color from the item to be inspected that is determined to be good, and A three-dimensional alignment processing unit applies a coordinate transformation mask generated from the shooting position and shooting angle when the inspection target image was captured to the feature data, and integrates multiple detection features from the multiple feature data generated corresponding to the multiple inspection target images included in the input image set, where the three-dimensional position of the inspection target is the same, to generate a three-dimensional feature map. A quality control device having the following features.

9. A feature extraction process takes an input image set containing multiple images of an item to be inspected, taken from multiple different angles, as input, and when the item to be inspected is determined to be good, the feature quantity output is set as the reference feature quantity, and the feature quantity output when the item to be inspected is determined to be good is set as the reference feature quantity, and the feature quantity data output includes detection feature quantities where the difference from the reference feature quantity is larger for parts of the item to be inspected that differ in shape and color from the item to be inspected that is determined to be good, and A three-dimensional alignment process is performed to generate a three-dimensional feature map by applying a coordinate transformation mask generated from the shooting position and shooting angle when the inspection target image was captured to the feature data, and integrating multiple detected features from the multiple feature data generated corresponding to the multiple inspection target images included in the input image set, where the three-dimensional position of the inspection target is the same, and generating a three-dimensional feature map. A quality control method that is performed by automated processing using a computer.

Images