JP5595247B2 - Mahalanobis reference space generation method and inspection device - Google Patents

Description

  The present invention relates to a technique for inspecting the appearance of a product using the Mahalanobis Taguchi method (MT method).

  Conventionally, the appearance inspection of products is largely dependent on human hands, and in this case, the inspector performs the inspection visually. However, in recent years, there has been a demand for automation of visual inspection of products using image recognition technology in response to demands for improving inspection quality from end users and measures for increasing labor costs due to human tactical inspection. An inspection apparatus that automatically performs an appearance inspection first images the appearance of an inspection product using an imaging device such as a camera or illumination. Then, the quality of the inspection product is determined by an algorithm that performs predetermined image processing on the captured inspection image. Here, a method using the MT method is conventionally known as one of the algorithms (see, for example, Patent Document 1).

  This MT method is a method in which a large number of pieces of information are collected on one scale called Mahalanobis distance, and the quality is determined based on the magnitude of the Mahalanobis distance. Specifically, a feature amount is extracted from a non-defective image obtained by capturing the appearance of a non-defective product, and a reference space (hereinafter also referred to as a Mahalanobis reference space) for calculating the Mahalanobis distance is generated. Next, the Mahalanobis distance is calculated based on the feature amount of the inspection image obtained by imaging the appearance of the inspection product and the reference space. The quality of the inspection product is determined based on the Mahalanobis distance.

  Here, a conventional Mahalanobis reference space generation method will be described. FIG. 16 is a flowchart showing a conventional procedure for generating a Mahalanobis reference space. First, in order to generate a Mahalanobis reference space, a normal image sample (non-defective image) serving as a reference for evaluation is acquired (S101). Next, the sample image acquired in S101 is cut into a predetermined size (for example, 640 × 480) (S102). Next, for the cropped image, 16 levels (R to 0 to 15, 16 to 31, 32 to 47,..., 240 to 255) are obtained from the R, G, and B brightness (8 bits: 0 to 255) data. , G, and B are counted in 48 steps), and the R, G, and B luminance distribution data are calculated (S103). Next, the luminance distribution data calculated in S103 is used as a feature amount of a non-defective image, and a Mahalanobis reference space is generated based on the feature amount (S104).

JP-A-2005-252451

  By the way, when the reference space generated as described above is used at the time of inspection, the inspection image to be inspected is also subjected to the cutting process as in the above-described S102, and the brightness of each cut image is determined. It is necessary to obtain distribution data. However, when the range of the clipped image becomes a large range such as 640 × 480, an operation is required for 640 × 480 = 307200 pixels. Furthermore, since this method needs to be performed with the size of the inspection image (frame size), the number of images to be cut increases and the calculation time further increases. Although it is necessary to perform numerical calculations in the MT method, in recent CCD cameras that capture the appearance of products, what can be captured in units of millions of pixels is becoming more and more versatile due to the demand for higher resolution. In order to process the pixels, the above-described conventional method has a problem that the calculation amount increases, and as a result, the inspection time becomes long.

In the conventional method, an image may appear unexpectedly at the time of inspection. Specifically, for example, there are cases where the entire image appears dark or bright, or where a part of the image cut out by image processing is dark and part is bright. Note that the following (1) to (3) can be considered as causes for the image to appear dark or bright.
(1) When accidental due to disturbances, etc. (2) When lighting is used for a long period of time (due to deterioration of lighting) (3) Due to the influence of the surface shape and material of the inspection product, the surface of the inspection product is uniform. When it is difficult to irradiate light

  When an image is unexpectedly reflected in this way, a shift in the brightness of the entire image occurs regardless of whether or not an area considered to be defective is reflected in the image. As the luminance shifts, the luminance distribution data also shifts, so that the Mahalanobis distance (MD value) increases regardless of whether the image is good or defective. In this case, there is a problem that it leads to over-detection of the inspection device (determining a non-defective product as defective).

  The present invention has been made in view of the above problems, and relates to a technique for determining pass / fail of an inspection product using the MT method, wherein the pass / fail determination is performed with high accuracy regardless of whether the image is dark or bright. One issue. A second problem is to shorten the processing time for determining pass / fail.

In order to solve the above problems, the method for generating the Mahalanobis reference space according to the present invention is a non-defective image acquisition step for acquiring a non-defective image including a non-defective gray region, which is a region where it is difficult to distinguish whether it is a non-defective product,
A recognition step for recognizing the non-defective gray region included in the non-defective image acquired in the non-defective image acquisition step;
A luminance acquisition step of acquiring luminance information of a peripheral region that is a region of a predetermined size adjacent to an edge of the non- defective gray region recognized in the recognition step in the non-defective image region;
A feature amount calculating step for calculating a feature amount of the luminance information of the non-defective gray region based on the luminance information of the peripheral region acquired in the luminance acquiring step;
Generating a reference space for calculating a Mahalanobis distance based only on the feature amount of the non-defective gray region of the non-defective image .

  According to this, the reference space is generated by paying attention to the non-defective gray region included in the non-defective image. Here, at present, the inspection by the visual inspector does not determine the defect by looking only at the defective area, but distinguishes the defective area from the area around the defect relatively. That is, in order to determine that the target is definitely defective, information around the target is required. In this regard, in the present invention, since the feature amount of the luminance information of the non-defective gray region is calculated based on the luminance information of the peripheral region of the non-defective gray region, the feature amount considering the influence of the non-defective image reflection is calculated. Can be obtained. Since the reference space is generated based on the feature amount, it is possible to obtain a highly accurate reference space in consideration of the influence of the quality of the non-defective image. Therefore, by using this reference space when determining the quality of an inspection product, it is possible to accurately determine the quality regardless of how the image is reflected. In addition, since the reference space is generated from the non-defective gray area instead of the entire non-defective image, the processing time for generating the reference space and the processing time for determining the quality of the inspection product using the reference space Can be shortened.

Further, the feature amount calculating step in the present invention includes a representative value setting step of setting a representative value representative of luminance information of the peripheral area,
A luminance distribution calculating step of calculating, as the feature amount, luminance distribution data indicating a luminance information distribution of the non-defective gray region based on the representative value set in the representative value setting step. .

  According to this, since the representative value of the luminance information of the peripheral area is set in the representative value setting step, the luminance information of the peripheral area is reflected even when the luminance information of the peripheral area is distributed. Value (representative value) can be obtained. In the luminance distribution calculating step, the luminance distribution data based on the representative value is calculated, so that the feature amount of the non-defective gray region based on the luminance information of the peripheral region can be obtained.

  Here, the luminance distribution calculating step calculates a difference value between each of the luminance values greater than the representative value among the luminance values of the pixels in the non-defective gray region and the representative value, and calculates a total value of the difference values. The difference value between each of the luminance values smaller than the representative value and the representative value among the luminance values of the pixels in the non-defective gray region is calculated, and at least one of the total values of the difference values is calculated as the luminance distribution data. Step.

  According to this, in the luminance distribution calculation step, the total value of the difference values between the luminance values greater than the representative value and the representative value among the luminance values of the pixels in the non-defective gray region is calculated. A feature value indicating how bright the area is can be obtained. In addition, in the luminance distribution calculation step, instead of the total value or together with the total value, the difference value between each luminance value smaller than the representative value and the representative value among the luminance values of each pixel in the non-defective gray region is calculated. Since the total value is calculated, it is possible to obtain a feature amount indicating how dark the gray area is with respect to the surrounding area.

  The luminance distribution calculating step calculates, as the luminance distribution data, at least one of the number of pixels having a luminance value greater than the representative value and the number of pixels having a luminance value smaller than the representative value among the pixels of the non-defective gray region. It can also be a step to do.

  According to this, a feature amount indicating how much of the gray area is brighter than the surrounding area and a feature indicating how much of the area darker than the surrounding area is included. One or both of the quantities can be obtained.

  In addition, the luminance distribution calculating step includes the sum of the difference values between the luminance values greater than the representative value and the representative value among the luminance values of the pixels of the non-defective gray region, and the luminances of the pixels of the non-defective gray region. A total value of difference values between the luminance values smaller than the representative value and the representative value among the values, the number of pixels having a luminance value larger than the representative value among the pixels of the good gray region, and the good gray region In this step, at least one of the number of pixels having a luminance value smaller than the representative value among the respective pixels may be calculated as the luminance distribution data. Thereby, one or a plurality of feature amounts selected from the above four feature amounts can be obtained.

In the present invention, the feature amount calculating step includes:
A rearrangement step of rearranging the luminance value of each pixel of the non-defective gray region in the order of the luminance value;
And a change characteristic value calculating step of calculating a change characteristic value indicating a change characteristic between the luminance values rearranged in the rearrangement step as the feature amount.

  According to this, the luminance value of each pixel in the non-defective gray area is rearranged in order of the magnitude of the luminance value in the rearrangement step, and the change characteristic between the rearranged luminance values is shown in the change characteristic value calculation step. Since the change characteristic value is calculated, it is possible to obtain a feature amount indicating how the luminance of the non-defective gray region is changing. Thereby, a reference space that further reflects the characteristics of the non-defective gray region can be obtained.

  Here, the change characteristic value calculation step calculates an adjacent change value that is a difference value between two luminance values adjacent to each other from the luminance value rearranged in the rearrangement step, and the characteristic of the adjacent change value The step of calculating the adjacent change characteristic value indicating the change characteristic value can be used.

  As a result, it is possible to obtain a feature value indicating how much the brightness changes between two adjacent brightness values.

  The change characteristic value calculating step calculates an approximate expression approximating a change between the luminance values rearranged in the rearranging step, and calculates an approximate expression characteristic value indicating the characteristic of the approximate expression as the change characteristic value. It is good also as a step to do.

  According to this, since the approximate expression characteristic value calculated in the change characteristic value calculating step shows the characteristic of the approximate expression that approximates the change between the luminance values, how the luminance of the gray region is changed between the pixels. It is possible to obtain a feature amount indicating whether or not it is changing.

In the present invention, the representative value setting step is a step of setting a plurality of representative values having different values for each good gray region,
The luminance distribution calculating step is a step of calculating the luminance distribution data based on each representative value for each of the plurality of representative values.

  According to this, since a plurality of representative values having different values are set in the representative value setting step, it is possible to obtain a value (representative value) in which the luminance information of the peripheral area is further reflected for each non-defective gray area. it can. In the luminance distribution calculation step, luminance distribution data based on each representative value is calculated for each of the plurality of representative values, so that the luminance information of the peripheral area for each non-defective gray area is further reflected. Brightness distribution data can be obtained. Thereby, a reference space with higher accuracy can be obtained.

  In the present invention, the representative value setting step may be a step of setting the representative value based on an average value, a minimum value, a maximum value, a median value, or a deviation of the luminance values in the peripheral area.

  According to this, since the average value, the minimum value, the maximum value, the median value, or the deviation can be determined for the first time in consideration of a plurality of luminance values, the accuracy can be improved by setting the representative value based on them. A good representative value can be obtained.

An inspection apparatus according to the present invention includes a storage unit that stores the reference space generated by the Mahalanobis reference space generation method according to any one of claims 1 to 10;
Inspection image acquisition means for acquiring an inspection image which is an image of an inspection product;
A gray region determination unit that determines whether or not the inspection image acquired by the inspection image acquisition unit includes an inspection product gray region that is a region that is difficult to distinguish whether it is a non-defective product;
When the gray area determining means determines that the inspection product gray area is included , a peripheral area that is a predetermined size area adjacent to an edge of the inspection product gray area in the inspection image area. Inspection product brightness acquisition means for acquiring brightness information;
An inspection product feature amount calculating means for calculating an inspection product feature amount that is a feature amount of the luminance information of the inspection product gray region based on the luminance information of the peripheral region acquired by the inspection product luminance acquisition unit;
Mahalanobis distance calculating means for calculating a Mahalanobis distance based on the inspection product feature value calculated by the inspection product feature value calculating means and the reference space stored in the storage means;
And a pass / fail judgment means for judging pass / fail of the inspection product based on the Mahalanobis distance calculated by the Mahalanobis distance calculation means.

  According to this, in order to determine the non-defective product of the inspection product, the Mahalanobis distance of the inspection product gray region that is difficult to distinguish whether it is non-defective product is calculated, so compared to the case of calculating the Mahalanobis distance of the entire inspection image, Processing time can be shortened. In addition, since the feature value of the brightness information of the inspection product gray area is calculated based on the brightness information of the peripheral area of the inspection product gray area, a highly accurate feature value that takes into account the influence of the appearance of the inspection image is taken into account. Can be obtained. The Mahalanobis distance is calculated based on the above-mentioned reference space in which the influence of the quality of the non-defective image is taken into consideration and the feature amount in which the influence of the appearance of the inspection image is taken into account. An accurate Mahalanobis distance that is not affected can be obtained. As a result, it is possible to accurately determine the quality of the inspection product regardless of the image appearance.

  Further, the Mahalanobis distance calculating means in the present invention includes a normalizing means for normalizing the inspection product feature value, and the Mahalanobis distance based on the inspection product feature value and the reference space normalized by the normalization means. It is characterized by calculating a distance.

  According to this, the normalization means normalizes the inspection product feature value, and the Mahalanobis distance calculation means calculates the Mahalanobis distance using the normalized inspection product feature value, so that an accurate Mahalanobis distance is obtained. be able to.

  The inspection apparatus of the present invention is characterized by comprising non-defective product determining means for determining that the inspection product is a non-defective product when the gray region determining device determines that the inspection product gray region is not included.

  According to this, when the inspection image does not include the inspection product gray area, the non-defective product determination unit determines that the inspection product is a non-defective product, and thus the process of calculating the Mahalanobis distance for the inspection image is omitted. can do. Therefore, it is possible to shorten the processing time for determining the quality of the inspection product.

1 is a diagram illustrating an overall configuration of an inspection apparatus 1. FIG. It is the flowchart which showed the production | generation procedure of Mahalanobis reference space. It is the flowchart which showed the detailed procedure of S2 of FIG. It is the figure which illustrated the output result of the image processing of FIG. It is the flowchart which showed the detailed procedure of S3 of FIG. It is the flowchart which showed the detailed procedure of S31 of FIG. FIG. 7 is a diagram illustrating a non-defective image 62 in the process of each processing step in FIG. 6. It is the flowchart which showed the detailed procedure of S32 of FIG. FIG. 6 is a diagram showing a non-defective gray region 621. It is a figure explaining the calculation method of the integral characteristic of S322 of FIG. 8, and the pixel number characteristic of S323. It is the flowchart which showed the detailed procedure of S325 of FIG. It is the flowchart which showed the detailed procedure of S326 of FIG. It is the flowchart which showed the detailed procedure of S4 of FIG. It is the flowchart which showed the quality determination procedure of the inspection goods. It is the figure which showed the images 71-79 of the some sample AI of a test | inspection goods. It is the flowchart which showed the production | generation procedure of the conventional Mahalanobis reference space.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments of a method for generating a Mahalanobis reference space and an inspection apparatus according to the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating an overall configuration of an inspection apparatus 1 according to the present embodiment. The inspection device 1 is a device that determines the quality of the appearance of the product by inspecting the surface of the product such as a cut metal part for unevenness, dust, scratches, burrs, and the like. As shown in FIG. 1, the inspection apparatus 1 includes an image processing PC 10, a notification device 20, a camera 30, and an inspection table 50. The inspection table 50 is a table on which a non-defective product sample 60 (see FIG. 1) for generating the Mahalanobis reference space and an inspection product 70 (see FIG. 1) to be inspected are placed. The camera 30 has a lens 31, and images the non-defective sample 60 and the inspection product 70 placed on the inspection table 50 through the lens 31. As the camera 30, for example, a CCD image sensor or a digital still camera can be used. In order to allow the camera 30 to image the good product sample 60 and the inspection product 70 satisfactorily, illumination (not shown) for irradiating the good product sample 60 and the inspection product 70 with light is appropriately provided.

  An image captured by the camera 30 is transmitted to the image processing PC 10 via the cable 40 (see FIG. 1). The image processing PC 10 is a computer that performs predetermined image processing on an inspection image, which is an image of the inspection product 70, and uses the MT method to determine non-defectiveness of the inspection image, that is, pass / fail determination of the inspection product 70. is there. The image processing PC 10 is also a computer that generates a Mahalanobis reference space used in the MT method based on a non-defective image, which is an image of the non-defective sample 60, prior to determining the quality of an inspection image. The computer that inspects the inspection product 70 and the computer that generates the Mahalanobis reference space may be separated.

  More specifically, the image processing PC 10 includes a CPU 11 and a memory 12. The CPU 11 executes a process according to a control program stored in the memory 12, thereby generating a Mahalanobis reference space or performing an inspection. The quality of the product 70 is judged. Details thereof will be described later. The memory 12 is a storage device such as a ROM or a flash memory in which a control program executed by the CPU 11 is stored. In the present embodiment, the memory 12 also stores the Mahalanobis reference space generated by the image processing PC 10.

  The notification device 20 notifies an inspection result (a result of quality determination of the inspection product 70) performed by the image processing PC 10. As the notification device 20, for example, a speaker for notification by sound or a display for notification by display can be used.

  Next, a method for generating the Mahalanobis reference space executed by the image processing PC 10 will be described in detail. The timing at which the image processing PC 10 generates the Mahalanobis reference space may be any timing as long as the inspection product 70 is not inspected. Here, FIG. 2 is a flowchart showing a procedure for generating the Mahalanobis reference space. First, the procedure for generating the Mahalanobis reference space will be roughly described with reference to FIG. 2, and then the details of each procedure in FIG. 2 will be described. Note that the processing in FIG. 2 is started, for example, when a predetermined start switch (not shown) is operated by the user.

  In FIG. 2, first, a non-defective product sample that is known as a non-defective product, that is, a non-defective sample that the visual inspector has determined to be non-defective is placed on the inspection table 50 (see FIG. 1). Then, the image (good product image) is acquired (collected) (S1). At this time, non-defective images of a plurality of non-defective samples are acquired. Next, predetermined image processing is performed on each non-defective image acquired in S1, and a non-defective gray region that is difficult to distinguish whether it is non-defective or not included in each non-defective image is detected (S2). ). Next, the feature amount of the luminance information of the detected non-defective gray region is calculated (S3). Next, a Mahalanobis reference space is generated based on the feature amount calculated in S3 (S4). Thereafter, the processing of the flowchart of FIG.

  As described above, in the present invention, the Mahalanobis reference space is not generated from the entire non-defective image, but is generated from the non-defective gray region of the non-defective image. However, the Mahalanobis reference space is not generated considering only the non-defective gray region, and the Mahalanobis reference space is generated considering the peripheral region of the good gray region. Hereinafter, each procedure of FIG. 2 will be described in detail.

  FIG. 3 is a flowchart showing a detailed procedure of the image processing (S2) for detecting the non-defective gray region in FIG. Specifically, the image processing is performed in the order of binarization processing (S21), smoothing processing (S22), and difference processing (S23) on the non-defective product image acquired in S1 of FIG. The binarization process, the smoothing process, and the difference process are general enough that those skilled in the art do not need to explain them, so the detailed description of each process is omitted. The region detected by the image processing of S21 to S23 is a non-defective gray region that is difficult to distinguish whether it is a non-defective product. Next, both the non-defective gray region and the original non-defective image before the image processing of S21 to S23 is output (S24). Then, the process of the flowchart of FIG. 3 is complete | finished.

  The purpose of the image processing in FIG. 3 is to perform processing so that almost all defects can be detected. In other words, in general, the overdetection rate and the defect outflow rate are in a trade-off relationship, but since the image processing here may perform overdetection, Image processing is performed to detect. That is, the non-defective gray region here is the one having the possibility of the defect even a little. In addition, in order to be able to detect almost all of the defects, samples that are known to be defective, which are determined in advance by a visual inspector, are collected, and image processing is performed so that all of them can be detected, that is, each image processing of S21 to S23. (For example, a threshold value when performing binarization processing) is determined.

  It should be noted that the image processing for detecting the non-defective gray region is performed as much as possible out of the existing algorithms, regardless of the algorithm of the processing procedure of FIG. 3 (binarization processing → smoothing processing → difference processing). Any algorithm that is configured to prevent may be used by those skilled in the art.

  Here, FIG. 4 is a diagram illustrating an output result of the image processing of FIG. 3, and more strictly, input images 61 to 65 of a plurality of non-defective samples A to E and outputs to these input images 61 to 65. It shows. Since the binarization process, the smoothing process, and the difference process are general enough that those skilled in the art need not to explain them, the intermediate process of the above process is omitted in FIG. As shown in FIG. 4, here, five good samples A to E are prepared, and good images 61 to 65 of these good samples A to E are acquired in S1 of FIG. Then, the non-defective images 61 to 65 are input images for performing the image processing of FIG. The good image 61 of the non-defective sample A and the non-defective image 64 of the good sample D are assumed to be clean images with no dirt or unevenness, and the good sample B is assumed to have unevenness on the surface. Therefore, the non-defective product image 62 includes a non-defective product gray region 621 showing unevenness. Similarly, it is assumed that the non-defective product sample E has unevenness on the surface thereof, and therefore, the non-defective product image 65 includes a non-defective product gray region 651 showing unevenness. Further, it is assumed that the non-defective sample C has dust on its surface, and therefore the non-defective image 63 includes a non-defective gray region 631 showing dust.

  By performing the image processing of S21 to S23 in FIG. 3 on these non-defective images 61 to 65, the output image in the lower column of FIG. 4 is output in S24. Specifically, for the good product image 62 of the good product sample B, the good product gray area 621 and the good product image 62 included in the good product image 62 are output. For the good product image 63 of the good product sample C, the good product gray region 631 and the good product image 63 included in the good product image 63 are output. For the good product image 65 of the good product sample E, the good product gray region 651 and the good product image 65 included in the good product image 65 are output. Since the non-defective sample A and the non-defective sample D are non-defective products that do not include the non-defective gray region, nothing is output in S24 of FIG. Whether unevenness or dust is a good product or defective depends on the detection requirements, but in the example of FIG. 4, it is assumed that the unevenness and dust are treated as good products as just noise. Therefore, in the image processing of S21 to S23 in FIG. 3, it is assumed that overdetection is performed on the non-defective images 62, 63, and 65.

  Next, details of the feature amount calculation processing in S3 of FIG. 2 will be described. Here, FIG. 5 is a flowchart showing a detailed procedure of the feature amount calculation processing. In the present invention, the luminance distribution data of the non-defective gray region is calculated as the feature amount of the non-defective gray region based on the luminance information of the peripheral region of the non-defective gray region. Specifically, first, in order to use the luminance information of the peripheral area of the non-defective gray area as a reference, a threshold value TH representing the luminance information of the peripheral area is set (S31). The threshold value TH corresponds to the “representative value” of the present invention. FIG. 6 is a flowchart showing a detailed procedure for setting the threshold. Here, the non-defective product image 62 and the non-defective gray region 621 of the sample B in FIG. 4 will be described as the processing target in FIG. FIG. 7 is a diagram showing a non-defective image 62 in the process of each processing step in FIG.

  The non-defective product gray area 621 detected in the previous S2 and the non-defective product image 62 (see FIG. 7A) including the same are input. First, the non-defective product gray area 621 included in the non-defective image 62 is enlarged. An area enlargement process is executed (S311). Specifically, as shown in FIG. 7B, for the non-defective gray region 621 detected in S2, the non-defective gray region 621 is enlarged by a specific range over the entire periphery of the edge 621b of the non-defective gray region 621. 622 is acquired from the non-defective image 62 which is the original image (S311). In the present embodiment, the enlargement ratio of the non-defective gray region 621 is set to the edge 621b + 1pixel (see FIG. 7B).

  Next, a difference process is performed to extract a difference area 623 between the non-defective gray area 621 and the enlarged area 622 (S312; see FIG. 7C). The difference area 623 can be extracted by performing (enlarged area 622) − (non-defective gray area 621). This difference area 623 is set as a peripheral area of the non-defective gray area 621. Hereinafter, the difference area 623 is referred to as a peripheral area.

  Next, the luminance value of each pixel in the peripheral area 623 is acquired (S313). Here, for the sake of brevity, a specific example of the processing of S313 will be described assuming that the number of pixels in the peripheral region 623 is 15 as shown in FIG. In S313, numbers are sequentially assigned to the respective pixels 623a in the peripheral area 623. Here, as shown in FIG. 7D, each pixel 623a has a smaller number as it goes to the upper left side of FIG. 7D, in other words, a higher number that goes to the lower right side of the page. Numbers are assigned in order from the first. In this case, numbers up to 15 are allocated. In parallel with assigning a number to each pixel 623a, the luminance value of each pixel 623a is acquired. The luminance data thus obtained is shown in Table 1 below. In Table 1, luminance values are shown in order of assigned numbers.

  Next, the minimum value of each luminance value in Table 1 (1 in the present embodiment: luminance value 35) and the average value of each luminance value are acquired (S314). However, since the luminance value is represented by an integer, the calculated average value (41.067) is rounded down to obtain 41 as the average luminance value.

  Next, a threshold value TH representing the luminance information of the peripheral area 623 is set based on the minimum value and the average value of the luminance values (S315). Specifically, a plurality of threshold values TH are set in a range between the obtained minimum value and average value. Here, the threshold TH for calculating the feature amount of the non-defective gray region 621 is set to 35, 37, 39, 41 in the range of 35 (minimum value) to 41 (average value). The threshold TH is set in the range between the minimum value and the average value. For example, the number of threshold values TH to be set is determined in advance. The value of each equally dividing point when the range is equally divided is set as the threshold value TH. Note that the non-defective gray regions 631 and 651 (see FIG. 4) other than the non-defective gray region 621 are also subjected to the same processing as described above, and the threshold value TH is set for each of the regions 631 and 651.

  In the present embodiment, the threshold value TH is set based on the minimum value and the average value of the luminance. However, for example, a deviation of the luminance value, a median value, a maximum value, and the like can be used. For example, when using the minimum value and the deviation, the threshold TH may be arbitrarily set within the range of the minimum value ± 3σ (σ = standard deviation), or the threshold TH may be set within the range of the maximum value ± 3σ. it can. As described above, there are any number of methods for setting the threshold value TH, but the feature of the present invention is that the threshold value TH is determined from the luminance information of the peripheral region of the non-defective gray region.

  After setting the threshold value TH in S315, the process of the flowchart of FIG. 6 is terminated, and the process returns to the process of FIG. Next, the luminance distribution data of the non-defective gray region is calculated based on the threshold value TH set in S31 (S32). FIG. 8 is a flowchart showing a detailed procedure for calculating the luminance distribution data. Here, the non-defective product image 62 of the sample B in FIG. 4 will be described as the processing target in FIG. Further, the description will be made assuming that the number of pixels of the non-defective gray region 621 included in the non-defective image 62 is 11 pixels. First, the luminance value of each pixel of the non-defective gray region 621 included in the non-defective image 62 is acquired (S321). Here, FIG. 9 is a diagram showing a non-defective gray region 621. In S321, numbers are sequentially assigned to the respective pixels 621a (see FIG. 9) in the non-defective gray region 621. Here, as in the case of the peripheral area described above, the numbers from 1 to 11 are assigned to each pixel 621a so that the number increases as it goes to the upper left in FIG. 9, in other words, to the lower right. Numbers are assigned in order. In parallel with assigning a number to each pixel 621a, the luminance value of each pixel 621a is acquired. The luminance data thus obtained is shown in Table 2 below. In Table 2, the luminance values are shown in the order of the assigned numbers.

  Next, based on each luminance value in Table 2 and the threshold value TH set in S31 of FIG. 5, the luminance distribution data of the non-defective gray region 621 with reference to the threshold value TH is calculated (S322, S323). Specifically, first, as the luminance distribution data of the non-defective gray region 621, an integral characteristic indicating how bright the non-defective gray region 621 is with respect to the peripheral region 623 (see FIG. 7) is calculated (S322). ). Here, FIG. 10A is a diagram for explaining a method of calculating the integral characteristic. Specifically, the horizontal axis represents a pixel number and the vertical axis represents a luminance value graph. In the graph of FIG. 10A, each luminance value in Table 2 is plotted with dots. Note that the points of the respective luminance values are indicated by P1 to P11. Further, in FIG. 10A, the line L1 with the threshold TH set earlier (the line L1 with the threshold TH = 35 in FIG. 10A) is drawn.

  In order to calculate the integral characteristic, a value obtained by subtracting the threshold value TH from the luminance values P3, P5, P6, P9, P10, and P11 higher than the line L1 of the threshold value TH among the luminance values P1 to P11 is set to each luminance value. Calculation is performed for each of P3, P5, P6, P9, P10, and P11. Assuming that the difference value between the luminance value P and the threshold value TH is ΔP, in FIG. 10A, the difference value ΔP3 of the pixel number 3, the difference value ΔP5 of the pixel number 5, the difference value ΔP6 of the pixel number 6, and the difference of the pixel number 9 A value ΔP9, a difference value ΔP10 of the pixel number 10 and a difference value ΔP11 of the pixel number 11 are shown. Note that the difference value ΔP for the luminance value smaller than the threshold value TH is calculated as ΔP = 0. Then, the total value of each difference value ΔP (= Σ (luminance value−threshold value TH)) is calculated as an integral characteristic. FIG. 10A shows an example in which the threshold value TH = 35, but in S322, the integral characteristics are calculated for other threshold values TH (= 37, 39, 41), respectively.

  In this description, the luminance value above (larger) the threshold value TH is targeted, but conversely, the integral characteristic (= Σ (= (())) corresponding to the area of the luminance value below (smaller) the threshold value TH. Threshold value TH-luminance value)) may be calculated. In this case, it is possible to obtain a feature amount indicating how dark the non-defective gray area is with respect to the surrounding area. Further, a part of the plurality of threshold values TH may be targeted for the luminance value on the upper side, and another threshold value TH may be targeted for the luminance value on the lower side.

  Returning to the description of FIG. 8, after calculating the integral characteristics in S <b> 322, the area that is brighter than the peripheral area 623 (see FIG. 7) in the non-defective gray area 621 as the luminance distribution data of the non-defective gray area 621. The pixel number characteristic indicating how much is included is calculated (S323). Here, FIG. 10B is a diagram for explaining the calculation method of the pixel number characteristic. Specifically, as in FIG. 10A, the horizontal axis represents the pixel number and the vertical axis represents the luminance value. It is the figure which showed the line L1 of the points P1-P11 and threshold value TH which plotted each luminance value on the graph. As the pixel number characteristic, the number of pixels having the luminance value P equal to or higher than the threshold value TH is counted. As shown in FIG. 10B, the luminance values P equal to or higher than the threshold value TH are set as luminance values P3, P4, P5, P6, P9, P10, and P11. Therefore, when the threshold value TH = 35, the pixel number characteristic = 7 (pixels). FIG. 10B shows an example in which the threshold value TH = 35. However, in S323, the pixel number characteristics are calculated for other threshold values TH (= 37, 39, 41).

  Regarding the calculation of the pixel number characteristic, the number of pixels having a luminance value smaller than the threshold TH may be counted contrary to counting the number of pixels having a luminance value equal to or higher than the threshold TH. In this case, it is possible to obtain a feature amount indicating how much of the gray area is darker than the surrounding area. Further, a part of the plurality of threshold values TH may be targeted for pixels having a brightness value equal to or higher than the threshold value TH, and another threshold value TH may be targeted for pixels having a brightness value smaller than the threshold value TH. Note that either the process of S322 or the process of S323 may be executed first, or may be executed simultaneously in parallel.

  The luminance distribution data obtained in S322 and S323 is shown in Table 3 below. Table 3 shows the luminance distribution data of the non-defective gray region 621 that is the subject of the description. The threshold value TH shown in Table 3 is a value limited to the non-defective gray region 621. If the information on the non-defective gray region is changed, the luminance information of the peripheral region is changed accordingly, and thus the threshold value TH is changed. That is, the threshold value TH is set for each good gray region.

  Only the luminance distribution data of the non-defective gray area calculated in S322 and S323 is a sufficient condition for generating the Mahalanobis reference space, but in order to generate a more accurate Mahalanobis reference space, other feature amounts are calculated. You may do it. Therefore, after the process of S323 in FIG. 8, it is determined whether or not the quality determination accuracy of the Mahalanobis reference space is to be improved (S324). This determination is made, for example, by confirming with the user whether or not to improve the pass / fail judgment accuracy of the Mahalanobis reference space, and based on the confirmation result from the user. Here, when it is determined that the quality determination accuracy of the Mahalanobis reference space is not improved (S324: No), the processing of the flowchart of FIG. 8 is terminated. In this case, since no other feature amount is calculated, the processing time for generating the reference space can be shortened. On the other hand, if it is determined to improve the pass / fail determination accuracy of the Mahalanobis reference space (S324: Yes), the process proceeds to S325.

  In S325, a rearrangement process is performed in which the luminance values of the pixels in the non-defective gray area acquired in S321 are rearranged in the order of the luminance values. Here, FIG. 11 is a flowchart showing a detailed procedure of the reordering process of S325. Here, the case where the processing of FIG. 11 is applied to the luminance value of the non-defective gray region 621 in Table 2 will be described. First, the luminance values in Table 2 are sorted in ascending order and rearranged in ascending order (S331). The results at this time are shown in Table 4 below.

  Next, the pixel numbers are reset so that the pixel numbers of the respective pixels in Table 4 are in order (S332). The results at this time are shown in Table 5 below. Then, the process of the flowchart of FIG. 11 is complete | finished.

  Returning to the description of FIG. 8, next, in S326 to S328, a change characteristic value indicating a change characteristic between the luminance values rearranged in S325 is calculated. Specifically, first, an adjacent change characteristic value indicating a change characteristic between two luminance values whose sizes are adjacent to each other is calculated for the rearranged luminance values (S326). Here, FIG. 12 is a flowchart showing a detailed procedure of the processing of S326. In the process of this flowchart, i is 1, 2, and 3, and the loop process is performed three times. First, in order to initialize variables, n is set to 1 and counter is set to 0 (S341, S342). Next, the luminance value X (n) of the nth pixel is acquired from the data rearranged in S325 (see Table 5) (S343), and the luminance value X (n + 1) of the n + 1th pixel is acquired (S344). . However, n is 1, 2, 3, 4,..., Nmax (= the number of pixels in the non-defective gray region−1). In the case of Table 5, since 11 pixels, n is set to 1 to 10.

  Next, a difference value ΔX (= X (n + 1) −X (n)) between the luminance values X (n) and X (n + 1) is calculated (S345). Next, it is determined whether or not the difference value ΔX is greater than or equal to the variable i (S346). If the difference value ΔX is greater than or equal to the variable i (S346: Yes), 1 is added to the variable counter (S347). On the other hand, when the difference value ΔX is less than the variable i (S346: No), it is held in the current variable counter. Next, it is determined whether or not the variable n has reached Nmax (= the number of pixels in the non-defective gray region−1) (S348). If not reached yet (S348: No), 1 is added to n (S349), and the process returns to S343. In this case, the same processing as described above is repeated for adjacent luminance values X (n) and X (n + 1) whose pixel numbers are increased by 1 from the previous time (S343 to S348). In this way, until the variable n reaches Nmax, the number of difference values ΔX that are greater than or equal to the variable i is counted as the variable counter.

  If the variable n reaches Nmax in S348 (S348: Yes), the current value of counter is acquired (S350), and 1 is added to the variable i, and the above processing is performed for the new variable i. (S341 to S349) are repeated. When the variable i reaches Nmax when the variable i = 3 (S348: Yes, S350), the process of the flowchart of FIG.

  In the present embodiment, the number of loops is three, but the number of loops is not particularly limited. Further, although the counter is incremented by 1 when the difference value ΔX is greater than or equal to the variable i (S346, S347), the threshold value of S346 is not limited to the variable i or more. For example, the variable i × 2 may be set as a threshold, or a numerical value such as 1, 3, 5, 7 may be set as a threshold regardless of the variable i.

  Here, the processing of FIG. 12 will be described using specific numerical values in the case of the variable i = 1. In addition, it demonstrates using the numerical value of said Table 5 here. In Table 5, when the variable n = 1, since the first luminance value X (n) = 30 and the second luminance value X (n + 1) = 31, the difference value ΔX is calculated as 31-30 = 1. Is done. Since the difference value ΔX = 1 is 1 or more which is the value of the variable i, 1 is added to the counter. Thereafter, the variable n becomes 2, and the second luminance value X (n) = 31 and the third luminance value X (n + 1) = 31, so that the difference value ΔX = 0 is calculated. Therefore, the value of counter is retained and 1 is added to the variable n. This process is repeated, and at the stage when the variable n finally becomes 10, the process for the variable i = 1 is completed. Table 6 below shows the counter values when the variable i is 1, 2, and 3 for the non-defective gray region 621.

  Thus, in the process of S326, the value of counter is calculated as the adjacent change characteristic value. This adjacent change characteristic value is a relatively gentle change in luminance (minimum and maximum values of luminance) between a relatively steeply changing luminance (a large difference between the minimum and maximum luminance values) and non-defective noise. It plays the role of extracting the difference (small difference).

  Returning to the description of FIG. 8, after the process of S326, approximate expression characteristic values indicating characteristics of the approximate expression when the luminance values rearranged in S325 are expressed by an approximate expression are calculated (S327, S328). In the present embodiment, when the luminance values are approximated by a line of y = ax + b by the method of least squares, the slope a and intercept b of the line are calculated as approximate expression characteristic values. In the case of n luminance values (x1, y1), (x2, y2),..., (xn, yn) of n (in Table 5, n = 11), first, the slope a is calculated by the following equation 1. Calculate (S327). Note that n is the number of pixels, x is a pixel number, and y is a luminance value.

  Next, the intercept b is calculated by the following equation 2 (S328). Thereafter, the process of the flowchart of FIG.

  The slope a and the intercept b in the case of the non-defective gray region 621 (in the case of Table 5) are shown in Table 7 below.

  As described above, since the slope a and the intercept b indicate characteristics of an approximate expression that approximates the luminance value of each pixel, the characteristics of how the luminance of the non-defective gray region changes between the pixels. The quantity can be obtained. Note that the approximation formulas in S327 and S328 do not have to be linear approximation. For example, the slope a or the intercept b as a logarithmic function may be obtained as a feature quantity by using x in y = ax + b as a logarithm.

  Table 8 below shows the result of the luminance distribution data of the non-defective gray region 621 calculated by the processing of FIG. 8 described above.

  Hereinafter, in this specification, integration characteristics, pixel number characteristics, luminance changes, and the like are referred to as “items”, and luminance distribution data obtained from each item is collectively referred to as “features”. In Table 8, since the threshold value TH set in the items of the integral characteristic and the pixel number characteristic is for the non-defective gray region 621, the feature amounts of the other non-defective gray regions 631 and 651 (see FIG. 4) are calculated. In this case, a threshold value TH is set for each good gray area.

  Returning to FIG. 5, after calculating the luminance distribution data in S <b> 32, the process of the flowchart of FIG. 5 is terminated and the process returns to the process of the flowchart of FIG. 2. In FIG. 2, next, a Mahalanobis reference space is generated based on the feature amount calculated in S3 (S4). Here, FIG. 13 is a flowchart showing a detailed procedure of the process of S4. Hereinafter, in order to simplify the description, the method of generating the Mahalanobis reference space will be described by replacing the feature amount calculated in S3 with the symbols in Table 9 below. In Table 9, for each of the non-defective gray regions 621, 631, and 651 in FIG. In addition, the subscript n of each feature quantity X indicates the number of samples in the non-defective gray region, and the subscript k indicates the number of items.

  Several methods have been established as a method for generating the Mahalanobis reference space, but in the present embodiment, generation is performed by a frequently used calculation method. In the generation process, the relationship between n and k in Table 9 needs to be at least n> = k. In Table 9, the feature amount X is shown only for the non-defective gray regions 621, 631, and 651, but in fact, many other sample numbers n are secured. In the case of Table 9, the threshold value TH is determined for each of the non-defective gray regions 621, 631, and 651 in the calculation of the integral characteristic and the pixel number characteristic. However, for the sake of simplification, the threshold value TH, the luminance change, The explanation will be made with the inclination and intercept omitted.

  In FIG. 13, first, the feature quantity X is normalized by the following expression 3 (S41). In Equation 3, Xj (with overbar in Equation 3) is the average value of the jth item, and σj is the standard deviation of the jth item.

  Data x after normalization with Equation 3 is shown in Table 10 below.

  Next, the correlation coefficient rpq is calculated by the following equation 4 (S42). Thereafter, a correlation matrix R composed of the correlation coefficients r11 to rkk is generated (S42).

  Next, as shown in the following Expression 5, an inverse matrix A of the correlation matrix R is calculated (S43). Let the inverse matrix A be the Mahalanobis reference space.

  Thereafter, the process of the flowchart of FIG. 13 is terminated, and the process returns to the process of the flowchart of FIG. Then, after the process of S4, the process of the flowchart of FIG. As described above, the Mahalanobis reference space in the present invention is generated. The generated Mahalanobis reference space is stored in the memory 12 (see FIG. 1) for use in determining whether the inspection product is good or bad.

  Next, a method for determining pass / fail of an inspection product executed by the image processing PC 10 will be described in detail. FIG. 14 is a flowchart showing the procedure of the method. Note that the processing of the flowchart of FIG. 14 is started, for example, when a predetermined start switch (not shown) is operated by the user.

  First, an inspection product placed on the inspection table 50 (see FIG. 1) is imaged by the camera 30, and an inspection image which is a captured image is acquired (S51). Next, predetermined image processing is performed on the inspection image to detect a gray region (inspection product gray region) included in the inspection image (S52). The details of the process of S52 are the same as the process (process of FIG. 3) when generating the reference space described above. Further, the conditions of the image processing (binarization processing, smoothing processing, difference processing in FIG. 3) at this time are set to be the same as the conditions when the non-defective gray region of the non-defective image is detected. That is, the inspection product gray region is a region where it is difficult to distinguish whether it is a non-defective product, in other words, a non-defective product gray region.

  Next, it is determined whether or not an inspection product gray area is detected by the process of S52 (S53). This is because when the inspection product gray area is detected by the processing of S52, the inspection product gray area is output, and therefore the determination of S53 is performed based on whether or not the inspection product gray area is output. If the inspection product gray area is not detected (S53: No), it is determined that the inspection product is a non-defective product (S61), and the flowchart of FIG. 14 ends. On the other hand, when the inspection product gray area is detected (S53: Yes), the process proceeds to S54.

  In S54, luminance distribution data (feature amount) of the inspection product gray area is calculated (S54). The details of the process of S54 are the same as the process (S3 of FIG. 2) for generating the reference space described above. That is, luminance distribution data based on the peripheral region of the inspection product gray region is calculated. Next, the feature quantity, which is the luminance distribution data, is normalized by the same expression as Expression 3 above (S55). As the average value and standard deviation used in normalization, the average value Xj (with overbar in Expression 3) and the standard deviation σj of Expression 3 are used. Next, the Mahalanobis reference space stored in the memory 12 (see FIG. 1) is read (S56).

  Next, based on the Mahalanobis reference space, the Mahalanobis distance (MD value) of the inspection image, strictly speaking, the gray area of the inspection product included in the inspection image is calculated (S57). Specifically, the Mahalanobis reference space (see Equation 5 above) and the feature data normalized in S55 (x11 to xik, where k is the number of items, i is i = 1, 2, 3,... The MD value is calculated by substituting (the number of inspection product gray regions to be identified) into the following expressions 6 and 7.

  For example, if five inspection product gray areas are detected, when the number from i = 1 to 5 is assigned to each inspection product gray area, the number 1 (i = 1) to five MD values in total from the MD value in No. 5 to the MD value in No. 5 (i = 5).

  Next, it is determined whether or not the MD value is greater than or equal to a preset threshold value (S58). If the MD value is smaller than the threshold value (S58: No), it is determined that the inspection product is a non-defective product (S60). On the other hand, if the MD value is equal to or greater than the threshold value (S58: Yes), it is determined that the inspection product is defective (S59). There are several methods for setting the threshold, and one example is given below. That is, the MD value of the object to be inspected, which is known to be a good product or a defective product, is obtained, a test threshold value is set, and a loss in the case of erroneous determination is obtained. When a plurality of test threshold values are changed, the threshold value with the smallest loss is set as the threshold value used in S58. In order to prevent defective outflow as much as possible, a plurality of MD values to be inspected that are known to be defective in advance are obtained, and a threshold value is set with a value smaller than the minimum value of the obtained MD values.

  Thereafter, the notification device 20 (display or the like) notifies whether the product is good (S60) or defective (S59) (S62). Particularly in the case of defect determination, the degree of notification is increased by emphasizing on the display or making a sound so that the inspector can recognize it (S62). Then, the process of the flowchart of FIG.

  Next, the case where the quality determination by the inspection apparatus 1 (refer FIG. 1) of this embodiment is performed is given, giving a sample of a specific inspection product. Here, FIG. 15 shows images 71 to 79 of a plurality of samples A to I of the inspection product. Samples A to F shown in the upper part of FIG. 15 are good samples that the visual inspector has determined to be good. However, the non-defective samples A, B, and D have unevenness, and the images 71, 72, and 74 include areas (inspected product gray areas) 711, 721, and 741 that show unevenness. In addition, dust is present in the non-defective sample C, and therefore the image 73 includes a region 731 (inspected product gray region) showing dust. Further, the non-defective samples E and F are clean non-defective products, and therefore, the images 75 and 76 do not include the inspection product gray region. On the other hand, samples G to I shown in the lower part of FIG. 15 are defective samples that the visual inspector determines to be defective. Each of the defective samples G to I has a defect that causes the defect. For this reason, the images 77 to 79 include areas (inspected product gray areas) 771 to 791 indicating the defects.

  Further, when obtaining these images 71 to 79, the time zone for imaging is changed. Specifically, the images 72, 75, and 79 of the samples B, E, and I are taken in the night time zone, and therefore, the images 72, 75, and 79 are darkened as a whole. Further, the images 74, 76, and 77 of the samples D, F, and G are taken in the morning time zone, and for this reason, the images 74, 76, and 77 are brightened as a whole. In addition, the images 71, 73, and 78 of the samples A, C, and H are taken during the daytime period. For this reason, the images 71, 73, and 78 have normal brightness (the brightness at the time of imaging in the morning and the night time). (Brightness at the time of imaging).

  Table 11 below shows an example of the results of the pass / fail determination performed on the images 71 to 79 by the inspection apparatus 1.

  As shown in Table 11, since the images 75 and 76 (non-defective samples E and F) do not include the inspection product gray region, MD values are not calculated for the images 75 and 76. In addition, MD values of 1.2 to 1.8 are calculated for the non-defective samples A, B, C, and D including the inspection product gray region. In addition, MD values of 14.4 to 23.5 are calculated for the images 77 to 79 of the defective samples G to I. As described above, the inspection apparatus 1 sets the threshold value for obtaining the feature value of the gray area detected by the image processing by acquiring the luminance information of the peripheral area that changes depending on how the image is captured (brightness and darkness). The feature amount can be calculated without being affected by how the image is captured. Therefore, a sufficient difference in MD value can be obtained for the non-defective product and the defective MD value.

  As described above, in the present embodiment, the quality of the inspection product is determined based on the MD value of the inspection product gray area, which is difficult to distinguish whether it is a non-defective product, not the entire image. The processing time for performing can be shortened. Further, when the inspection product does not include the inspection product gray area, it is determined as a good product without calculating the MD value, so that the determination can be made quickly. Also, since the feature value of the gray area is obtained based on the luminance information of the peripheral area of the gray area (non-defective gray area, inspection gray area), the Mahalanobis reference space and MD values that are not affected by the way the image is captured are obtained. Can be calculated. Therefore, the quality determination can be performed with high accuracy regardless of the way the image is captured.

  In the above embodiment, the process of S1 in FIG. 2 corresponds to the “non-defective image acquisition step” of the present invention. The process of S2 in FIG. 2 corresponds to the “recognition step” of the present invention. The processing of S311 to S313 in FIG. 6 corresponds to the “luminance acquisition step” of the present invention. The process of S3 in FIG. 2 corresponds to the “feature amount calculating step” of the present invention. The process of S4 in FIG. 2 corresponds to the “generation step” of the present invention. The processes of S314 and S315 in FIG. 6 correspond to the “representative value setting step” of the present invention. The processing of S321 to S323 in FIG. 8 corresponds to the “luminance distribution calculating step” of the present invention. The process of S325 in FIG. 8 corresponds to the “rearrangement step” of the present invention. The process of S326 to S328 in FIG. 8 corresponds to the “change characteristic value calculation step” of the present invention. The memory 12 corresponds to the “storage means” of the present invention. The image processing PC 10 that executes the process of S51 of FIG. 14 corresponds to the “inspection image acquisition unit” of the present invention. The image processing PC 10 that executes the processing of S52 and S53 in FIG. 14 corresponds to the “gray area determination means” of the present invention. The image processing PC 10 that executes the process of S54 in FIG. 14 corresponds to the “inspection product luminance acquisition unit” and the “inspection product feature value calculation unit” of the present invention. The image processing PC 10 that executes the processes of S55 to S57 in FIG. 14 corresponds to the “Mahalanobis distance calculating means” of the present invention. The image processing PC 10 that executes the process of S55 of FIG. 14 corresponds to the “normalization means” of the present invention. The image processing PC 10 that executes the processes of S58 to S60 in FIG. 14 corresponds to the “good / bad determination means” of the present invention. The image processing PC 10 that executes the process of S61 in FIG. 14 corresponds to the “non-defective product determination unit” of the present invention.

1 Inspection device 10 Image processing PC
11 CPU
12 Memory 20 Alarm 30 Camera 61-65 Good product image 621, 631, 651 Good product gray region 623 Peripheral region 71-79 Inspection image 711, 721, 731, 741, 771, 781, 791 Inspection product gray region

Claims (13)

A non-defective image acquisition step for acquiring a non-defective image including a non-defective gray region, which is an area where it is difficult to distinguish whether or not it is a non-defective product,
A recognition step for recognizing the non-defective gray region included in the non-defective image acquired in the non-defective image acquisition step;
A luminance acquisition step of acquiring luminance information of a peripheral region that is a region of a predetermined size adjacent to an edge of the non- defective gray region recognized in the recognition step in the non-defective image region;
A feature amount calculating step for calculating a feature amount of the luminance information of the non-defective gray region based on the luminance information of the peripheral region acquired in the luminance acquiring step;
A generating step of generating a reference space for calculating a Mahalanobis distance based only on the feature quantity of the non-defective gray region in the non-defective product image . The feature amount calculating step includes a representative value setting step of setting a representative value representing luminance information of the peripheral area;
A luminance distribution calculating step of calculating, as the feature amount, luminance distribution data indicating a luminance information distribution of the non-defective gray region based on the representative value set in the representative value setting step. The method for generating the Mahalanobis reference space according to claim 1.   The luminance distribution calculating step calculates a difference value between each of the luminance values larger than the representative value and the representative value among the luminance values of the pixels of the non-defective gray region, and calculates a total value of the difference values and the good product Calculating a difference value between each luminance value smaller than the representative value among the luminance values of each pixel in the gray area and the representative value, and calculating at least one of the total values of the difference values as the luminance distribution data; The method of generating a Mahalanobis reference space according to claim 2, wherein:   In the luminance distribution calculating step, at least one of the number of pixels having a luminance value larger than the representative value and the number of pixels having a luminance value smaller than the representative value among the pixels of the non-defective gray region is calculated as the luminance distribution data. The method of generating a Mahalanobis reference space according to claim 2, wherein the method is a step.   The luminance distribution calculating step includes the sum of the difference values between the luminance values greater than the representative value and the representative values among the luminance values of the pixels in the good gray region, and the luminance values of the pixels in the good gray region. Of the luminance values smaller than the representative value and the difference value between the representative values, the number of pixels having a luminance value larger than the representative value among the pixels of the good gray region, and the good gray region 3. The method of generating a Mahalanobis reference space according to claim 2, wherein at least one of the number of pixels having a luminance value smaller than the representative value among the pixels is calculated as the luminance distribution data. The feature amount calculating step includes:
A rearrangement step of rearranging the luminance value of each pixel of the non-defective gray region in the order of the luminance value;
The change characteristic value calculation step of calculating a change characteristic value indicating a change characteristic between luminance values rearranged in the rearrangement step as the feature amount, further comprising: The method for generating the Mahalanobis reference space described in item 1.   The change characteristic value calculation step calculates an adjacent change value that is a difference value between two luminance values adjacent to each other from the luminance values rearranged in the rearrangement step, and shows the characteristics of the adjacent change value. The method of generating a Mahalanobis reference space according to claim 6, which is a step of calculating an adjacent change characteristic value as the change characteristic value.   The change characteristic value calculation step calculates an approximate expression that approximates a change between the luminance values rearranged in the rearrangement step, and calculates an approximate expression characteristic value indicating the characteristic of the approximate expression as the change characteristic value. The method of generating a Mahalanobis reference space according to claim 6 or 7, wherein the method is a step. The representative value setting step is a step of setting a plurality of representative values having different values.
9. The luminance distribution calculating step according to claim 2, wherein the luminance distribution calculating step is a step of calculating the luminance distribution data with reference to each representative value for each of the plurality of representative values. A method of generating the described Mahalanobis reference space.   10. The representative value setting step is a step of setting the representative value based on an average value, a minimum value, a maximum value, a median value, or a deviation of luminance values in the peripheral area. The method for generating the Mahalanobis reference space according to any one of the above items. Storage means for storing the reference space generated by the method for generating the Mahalanobis reference space according to any one of claims 1 to 10,
Inspection image acquisition means for acquiring an inspection image which is an image of an inspection product;
A gray region determination unit that determines whether or not the inspection image acquired by the inspection image acquisition unit includes an inspection product gray region that is a region that is difficult to distinguish whether it is a non-defective product;
When the gray area determining means determines that the inspection product gray area is included , a peripheral area that is a predetermined size area adjacent to an edge of the inspection product gray area in the inspection image area. Inspection product brightness acquisition means for acquiring brightness information;
An inspection product feature amount calculating means for calculating an inspection product feature amount that is a feature amount of the luminance information of the inspection product gray region based on the luminance information of the peripheral region acquired by the inspection product luminance acquisition unit;
Mahalanobis distance calculating means for calculating a Mahalanobis distance based on the inspection product feature value calculated by the inspection product feature value calculating means and the reference space stored in the storage means;
An inspection device, comprising: quality determination means for determining quality of the inspection product based on the Mahalanobis distance calculated by the Mahalanobis distance calculation means.   The Mahalanobis distance calculating means includes a normalizing means for normalizing the inspection product feature quantity, and calculates the Mahalanobis distance based on the inspection product feature quantity normalized by the normalization means and the reference space. The inspection apparatus according to claim 11, wherein the inspection apparatus is a thing.   13. The inspection apparatus according to claim 11, further comprising: a non-defective product determining unit that determines that the inspection product is a non-defective product when the gray region determining unit determines that the inspection product gray region is not included. .