JP2006266780A - Flaw inspection device and flaw inspection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately judge a flaw even in a restricted tact time. <P>SOLUTION: This flaw inspection device is constituted so that the surface of a test piece is photographed to acquire an original image (step A1), the average value of the brightness values of a noticiable pixel and its peripheral pixel is calculated on the basis of the original image and a filtered image wherein the calculated average value is replaced with the brightness value of the noticiable pixel is formed (step A2), the corresponding brightness value of the pixel of the filtered image is subtracted from the brightness value of the pixel of the original image to form a differential image (step A3), in a case that the brightness value of the pixel of the differential image is higher than a threshold value, the brightness value of the pixel is replaced with 1 and, when the brightness value of the pixel of the differential image is not higher than the threshold value, the brightness value of the pixel is replaced with zero to form a binary image (step A4) and the presence of a flaw is judged with respect to the binary image (steps A5, A6). <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a defect inspection apparatus and method for detecting defects on a cut surface of a casting, and more particularly to a defect inspection apparatus and method.

  Defects appearing on the machined surface of aluminum castings such as automobile parts are caused by grabbing the work piece while it is stuck to the chuck with indentations that do not know where they appear on the casting, or chips that cannot be removed by washing in the cutting machine. Indentation that occurs, and chatter that occurs due to variations in the rigidity of the workpiece. If these defects exist on the machined surface, the sealing performance is lowered and the product function is not satisfied. Therefore, in order to select castings that satisfy the product function, defects are inspected on the cut surfaces of the castings. Currently, all inspectors visually check the castings.

  In visual defect inspection, the inspection accuracy varies greatly depending on individual differences and the degree of fatigue due to work. In recent years, castings having a large shape and castings having a large number of cutting surfaces have been increased, which causes a large amount of expenses for inspection and increases manufacturing costs. Therefore, automation of defect inspection is desired.

  As a technique for automating defect inspection, there is a defect inspection apparatus using a camera and illumination. In such a defect inspection apparatus, as a countermeasure when a machining trace is generated due to various variation factors, the frequency component of the machining trace is specified by two-dimensional Fourier transform, and the machining trace is removed by removing the frequency component. Methods (Patent Document 1) and methods (Patent Documents 2 and 3) that absorb variations in measurement data by statistical processing have been proposed.

Japanese Patent Laid-Open No. 10-103938 JP 2004-170374 A JP 2004-264054 A

  However, in the invention described in Patent Document 1, since the frequency component of the machining trace is removed by two-dimensional Fourier transform, the scratch has the frequency component of the machining trace due to the deterioration of the device or the luminance change due to disturbance. If it occurs in a state, it may be removed in the same manner as the processing trace. In addition, when Fourier transform is used, the amount of calculation becomes very large, so that it is not suitable for processing with a limited tact time.

  Further, in the inventions described in Patent Documents 2 and 3, a plurality of non-defective workpieces are imaged, and a non-defective non-defective image is calculated using the average, standard deviation, or variance of the collected images, Since the created image is compared with the captured image and the defective product is judged, variations in the vicinity of the middle can be dealt with. In an image in a state where there is, a non-defective product is erroneously determined as a defective product. For example, when an LED element is used as illumination, stable determination performance can be ensured by making the brightness of the LED element saturated and reducing the influence of processing traces as much as possible. When it falls, a processing trace will generate | occur | produce and a non-defective product may be misjudged as a defective product. In addition, when the cutting tool is replaced, processing traces are likely to occur at the initial stage due to the shape of the cutting edge. Therefore, processing traces are generated even when the LED element uses a saturated luminance, and the non-defective product is similarly defective. May be misjudged. Further, in the inventions described in Patent Documents 2 and 3, statistical processing is performed on the N pieces of image data prepared, a threshold value for binarization processing is determined, and scratches are identified based on the determined threshold value. Therefore, in an image having a lot of processing marks, it is erroneously determined that there is a scratch even in a workpiece that does not have a scratch. Furthermore, in the inventions described in Patent Documents 2 and 3, it is necessary to prepare a large number of specimens in order to create a reference image, and the quality of selection of the specimens greatly affects the performance.

  An object of the present invention is to enable accurate determination even in a limited tact time.

  According to the inventor's view, robust performance and algorithms capable of high-speed computation are required to improve productivity in order to be able to cope with deterioration of equipment used and changes in luminance due to disturbance. . In addition, in order to be able to make an accurate determination even when the measurement environment changes extremely, such as variations in illumination or replacement of blades, a method using a threshold value that matches the state in which the environment has changed is necessary. From the above view, the following invention was made.

  According to a first aspect of the present invention, there is provided a defect inspection apparatus for inspecting a surface defect of a specimen, a camera that images the surface of the specimen and acquires an original image, and noise from the original image. A filter processing unit that creates a removed filtered image, a threshold value determining unit that determines a threshold value based on pixel information of the original image or pixels of the filtered image, and pixels of the filtered image When the luminance value of the pixel is greater than or equal to the threshold, the luminance value of the pixel is replaced with 1, and when the luminance value of the pixel of the filtered image is not equal to or greater than the threshold, the luminance value of the pixel is replaced with 0 The image processing apparatus includes: a binarization processing unit that creates a binarization processed image; and a determination processing unit that determines the presence / absence of a defect in the binarization processed image.

  In the defect inspection apparatus according to the present invention, the binarization processing unit includes a difference processing unit that creates a difference processed image by subtracting a luminance value of a pixel of the corresponding filtered image from a luminance value of the pixel of the original image. Replaces the luminance value of the pixel of the difference-processed image with 1 when the luminance value of the pixel of the difference-processed image is equal to or greater than the threshold value, and It is preferable to create a binarized image in which the luminance value is replaced with 0.

  In the defect inspection apparatus of the present invention, the threshold value determination unit may calculate an average value from luminance values of pixels of the original image or pixels of the filtered image and determine the threshold value.

  In the defect inspection apparatus of the present invention, the threshold value determination unit may determine a threshold value by calculating a variance value from luminance values of pixels of the original image or pixels of the filtered image.

  In the defect inspection apparatus of the present invention, the filter processing unit calculates an average value of luminance values of the target pixel and its surrounding pixels based on the original image, and calculates the calculated average value and the luminance value of the target pixel. An exchanged filtered image may be created.

  In the defect inspection apparatus of the present invention, the filter processing unit extracts the maximum value of the luminance value of the target pixel and its surrounding pixels based on the original image, and calculates the extracted maximum value and the luminance value of the target pixel. An exchanged filtered image may be created.

  In the defect inspection apparatus of the present invention, it is preferable that the defect inspection apparatus further includes an illumination unit that illuminates the surface of the specimen, and the threshold value determination unit determines the threshold value based on aged deterioration of the illumination unit.

  In a second aspect of the present invention, there is provided a defect inspection method for inspecting a surface defect of a specimen, the step of obtaining an original image by imaging the surface of the specimen, and noise from the original image. Creating a filtered filtered image; determining a threshold based on pixel information of the original image or pixel of the filtered image; and a luminance value of the pixel of the filtered image. When the luminance value of the pixel is not less than a threshold when the luminance value of the pixel of the filtered image is not equal to or greater than a threshold value, And creating and determining whether or not there is a defect in the binarized image.

  According to the present invention (claims 1-8), since a complicated process such as a two-dimensional Fourier transform is not performed, a non-defective product and a defective product can be accurately determined even in a limited tact time. In addition, it is not necessary to prepare a large number of specimens and create a reference image.

  According to the present invention (Claim 3-7), even when the illumination is deteriorated, the blade is replaced, etc., the threshold value is adjusted by the statistical method of the imaging screen to obtain stable determination performance.

(Embodiment 1)
A defect inspection apparatus according to Embodiment 1 of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram schematically showing a configuration of a defect inspection apparatus according to Embodiment 1 of the present invention.

  The defect inspection apparatus 1 is an apparatus that detects a defect (cast hole) on a cut surface of a casting. The defect inspection apparatus 1 includes a turntable 11, a motor 12, an encoder 13, a camera 14, an illumination 15, an image processing unit 16, and a display unit 17.

  The turntable 11 is a turnable table for placing the specimen 2 that is a casting. The motor 12 drives the turntable 11 to rotate. The encoder 13 is a part that measures the rotation angle of the motor 12. The camera 14 is a part that captures a surface image of the specimen 2 illuminated by the illumination 15, and a CCD camera, for example, can be used. The illumination 15 is a part that illuminates the surface of the specimen 2, and for example, an LED element can be used. The brightness of the illumination 15 is in a saturated state (high brightness). The image processing unit 16 performs defect inspection processing based on the image captured by the camera 14. The image processing unit 16 recognizes the position on the surface of the specimen 2 where the image captured by the camera 14 is based on the rotation angle of the motor 12 from the encoder 13. The display unit 17 displays a captured image, an image after the defect inspection process, and the like according to a user instruction. A generally used computer may be used for the image processing unit 16 and the display unit 17.

  Next, the processing operation of the image processing unit of the defect inspection apparatus according to the first embodiment will be described. FIG. 2 is a flowchart schematically showing the processing operation of the image processing unit of the defect inspection apparatus according to the first embodiment of the present invention. FIG. 3 is a flowchart schematically showing the binarization process of the image processing unit of the defect inspection apparatus according to the first embodiment of the present invention. 4A and 4B are images (images when the illumination brightness is 15) obtained by the defect inspection apparatus according to Embodiment 1 of the present invention, where FIG. 4A is an original image, FIG. 4B is a filtered image, and FIG. Is a difference processed image, and (d) is a binarized image.

  Each process shown in FIG. 2 and FIG. 3 is a process executed by a program in the image processing unit 16. Therefore, if each process is expressed by a functional block, it is equivalent to including an original image acquisition processing unit, a filter processing unit, a difference processing unit, a binarization processing unit, a feature amount calculation processing unit, and a determination processing unit. .

  First, the camera 14 captures an original image of the specimen 2 at high brightness (the illumination 15 is saturated), and the image processing unit 16 acquires the captured original image (see FIG. 4A) (step A1). ).

Next, the image processing unit 16 functioning as a filter processing unit performs a filter process on the original image in order to remove processing marks and scratches other than defects (ingots), and provides a filtered image (reference image; FIG. 4 (b)) is created (step A2). Here, in the filtering process, an average value of the luminance values of the target pixel and its peripheral pixels (n * n) is calculated, and the calculated average value and the luminance value of the target pixel are switched. In the filtering process, instead of using the average value, the maximum value is extracted from the luminance values of the target pixel and its surrounding pixels (n * n), and the extracted maximum value and the luminance value of the target pixel are switched. May be.

  Next, the image processing unit 16 functioning as a difference processing unit performs difference processing based on the original image and the filtered image, and creates a difference processed image (see FIG. 4C) (step A3). Here, the difference process is a process of subtracting the filtered image from the original image (subtracting the luminance value of the pixel of the corresponding filtered image from the luminance value of the pixel of the original image). If the original image has a scratch, the scratch portion appears as a difference in the difference processed image.

  Next, the image processing unit 16 functioning as a binarization processing unit performs binarization processing for each pixel on the difference processed image, and creates a binarized image (see FIG. 4D) (step A4). ). Here, in the binarization process, the image processing unit 16 extracts the luminance value (data (x, y)) of the pixel to be processed from the difference processed image (step B1), and the pixel luminance value and the threshold value. (Fixed value) is compared to determine whether the luminance value of the pixel is equal to or greater than a threshold value (step B2). If the luminance value of the pixel is equal to or greater than the threshold value (YES in step B2; In (part), the luminance value of the pixel is set to 1 (white) (step B3), and when the luminance value of the pixel is not equal to or higher than the threshold value (NO in step B2; a portion without scratches), the luminance value of the pixel is set to 0 (black) (Step B4), it is confirmed whether or not the binarization processing has been performed for all the pixels of the difference processed image (step B5), and when the binarization processing has not been performed for all the pixels (NO in step B5). Returns to step B1 to process the next pixel There, the process proceeds to step A5 when performing binarization processing for all the pixels (YES in step B5).

  Next, the image processing unit 16 calculates feature quantities such as the length and area of each scratched portion from the binarized image (step A5).

  Finally, the image processing unit 16 determines the presence or absence of a defect by comparing the feature amount with a threshold value (fixed value) (step A6). In the determination, if the feature amount is equal to or greater than the threshold value, the product is defective as a defect, and if the feature value is not equal to or greater than the threshold value, the product is determined to be non-defective. In the binarized image of FIG. 4 (d), there are three defects whose feature amount is equal to or greater than the threshold value, and it is determined as a defective product.

  According to the first embodiment, since a complicated process such as a two-dimensional Fourier transform is not performed, it is possible to accurately determine a non-defective product and a defective product even in a limited tact time.

(Embodiment 2)
Next, a defect inspection apparatus according to Embodiment 2 of the present invention will be described with reference to the drawings. FIG. 5 is a graph schematically showing a map (dispersion value-threshold value) used in the image processing unit of the defect inspection apparatus according to the second embodiment of the present invention. For the configuration of the defect inspection apparatus according to the second embodiment, refer to FIG.

  In the defect inspection apparatus according to the second embodiment, a dispersion value calculation process and a threshold correction process are added to the image processing unit 16 of the defect inspection apparatus according to the first embodiment. The configuration other than the image processing unit 16 is the same as that of the first embodiment.

  The image processing unit 16 calculates the total value, average value, or variance value of the luminance values of each pixel based on the original image of the specimen 2 captured by the camera 14 in the variance value calculation process. Here, the total value is a value obtained by summing the luminance values of the respective pixels of the original image. The average value is a value obtained by dividing the total value by the number of pixels of the processed surface of the original image. Furthermore, the variance value is a value calculated by the following arithmetic expression (Equation 1).

area:切削面の面積
x,y:画素位置（切削面のみ有効）
ave:原画像切削面の平均輝度
n,m:画像サイズ

area: cutting surface area
x, y: Pixel position (valid only on the cutting surface)
ave: Average brightness of the cut surface of the original image
n, m: Image size

  In the threshold value correction process, the image processing unit 16 calculates a threshold value used in the binarization process based on the calculated total value, average value, or variance value, and corrects the calculated threshold value. Here, since the detection of scratches increases as the variance value increases, the threshold value is set higher as the variance value increases (see FIG. 5). When using the total value or the average value, the threshold value is set lower as the total value or the average value becomes higher.

  Next, the relationship among the brightness, total value, variance value, and threshold value (correction value) of the illumination 15 of the defect inspection apparatus according to the second embodiment will be described with reference to the drawings. FIG. 6 is a graph for explaining the relationship between the luminance of illumination and the total value in the defect inspection apparatus according to Embodiment 2 of the present invention. FIG. 7 is a graph for explaining the relationship between the luminance of illumination and the dispersion value in the defect inspection apparatus according to Embodiment 2 of the present invention. FIG. 8 is a graph for explaining the relationship between the luminance of the illumination and the threshold value (correction value) in the defect inspection apparatus according to the second embodiment of the present invention.

  When the specimen 2 is imaged, if the brightness of the illumination 15 is reduced due to deterioration of the illumination 15 or the like, processing marks are likely to occur. Further, if the brightness of the illumination 15 is too high, small scratches may disappear. In the binarization processing, if a fixed threshold value is used as in the first embodiment, there is a possibility that the processing mark is determined to be a scratch when the luminance is low, and a small scratch may not be detected when the luminance is high. Further, since processing marks are likely to occur when the cutting tool is replaced, there is a possibility that the processing marks are erroneously determined as scratches. Therefore, for low-luminance images, it is necessary to accurately separate processing marks and scratches from image data in which processing marks and scratches are mixed. A high-luminance image can be easily determined because no processing marks are generated. When the illumination 15 is deteriorated, it is possible to avoid such an erroneous determination by replacing the illumination 15 with a new one. However, if the threshold value used for the binarization process is corrected based on some characteristic, an error may occur. It should be possible to reduce the number of replacements of the illumination 15 while avoiding the determination.

  Here, if statistical processing is performed on an image in which there is no processing mark and an image in which the processing mark is generated, there is a relationship between the average value of the luminance value in the detection range, the dispersion value, and the generation of the processing mark. When the illumination 15 is deteriorated and processing marks are generated, the average value of luminance is lowered and the dispersion value is increased. In addition, processing marks are generated even when the blade is replaced, and the average value of the luminance decreases, the luminance variation of the imaging surface increases, and the dispersion value tends to increase. When the luminance (for example, 15 levels) of the illumination 15 is changed, the total value of the luminance values of the pixels related to the image is as shown in FIG. That is, as the luminance of the illumination 15 increases (approaches the luminance 15), the total luminance value of each pixel related to the image increases. Further, when the luminance of the illumination 15 (for example, 15 levels) is changed, the dispersion value of the luminance value of each pixel related to the image is as shown in FIG. That is, as the luminance of the illumination 15 increases (approaches the luminance 15), the variance value of the luminance value of each pixel related to the image decreases. The total value or the variance value of the luminance values of each pixel related to the image has a proportional or inversely proportional relationship with the luminance of the illumination 15.

  When the average value of the brightness value of the cutting surface is reduced or the variance value is increased by utilizing the above characteristics, it is assumed that a lot of machining marks are generated, and the threshold value used for the binarization process. Correct the value. As for the relationship between the brightness of the illumination 15 and the threshold value (correction value), when the brightness is low, the threshold value is set high to detect machining marks and scratches when the threshold value is low. Since there are few machining marks, the threshold value is set low (see FIG. 8). As a result, it is possible to prevent the processing mark from being erroneously determined as a scratch and to detect a small scratch at high luminance.

  Next, the processing operation of the image processing unit of the defect inspection apparatus according to the second embodiment will be described. FIG. 9 is a flowchart schematically showing the processing operation of the image processing unit of the defect inspection apparatus according to the second embodiment of the present invention. FIG. 10 is a flowchart schematically showing the binarization process of the image processing unit of the defect inspection apparatus according to the second embodiment of the present invention. 11A and 11B are images (images when the illumination brightness is 1) obtained by the defect inspection apparatus according to the second embodiment of the present invention, where FIG. 11A is an original image, FIG. 11B is a filtered image, and FIG. Is a difference processed image, and (d) is a binarized image.

  Each process shown in FIGS. 9 and 10 is a process executed by a program in the image processing unit 16. Therefore, if each process is expressed by a functional block, an original image acquisition processing unit, a variance value calculation processing unit, a threshold value correction processing unit, a filter processing unit, a difference processing unit, a binarization processing unit, a feature amount calculation processing unit This is equivalent to having a determination processing unit.

  First, an original image of the specimen 2 illuminated by the illumination 15 is captured by the camera 14, and the image processing unit 16 acquires the captured original image (see FIG. 11A) (step C1).

  Next, the image processing unit 16 calculates a variance value based on the acquired original image (step C2). The calculation of the variance value is performed based on the arithmetic expression (Equation 1).

  Next, the image processing unit 16 functioning as a threshold value determination unit performs threshold value correction processing based on the calculated variance value (step C3). In the threshold value correction process, a threshold value used in a binarization process to be described later is calculated based on the calculated variance value, and is corrected to the calculated threshold value.

  Next, the image processing unit 16 performs filter processing on the original image in order to erase the processing marks and scratches, and obtains a filtered image (reference image; see FIG. 11B) (step C4). Here, in the filtering process, an average value of the luminance values of the target pixel and its peripheral pixels (n * n) is calculated, and the calculated average value and the luminance value of the target pixel are switched. In the filtering process, instead of using the average value, the maximum value is extracted from the luminance values of the target pixel and its surrounding pixels (n * n), and the extracted maximum value and the luminance value of the target pixel are switched. May be.

  Next, the image processing unit 16 performs difference processing based on the original image and the filtered image, and obtains a difference processed image (see FIG. 11C) (step C5). Here, the difference process is a process of subtracting the filtered image from the original image (subtracting the luminance value of the pixel of the corresponding filtered image from the luminance value of the pixel of the original image). If the original image has a scratch, the scratch portion appears as a difference in the difference processed image.

  Next, the image processing unit 16 performs a binarization process for each pixel on the difference processed image to obtain a binarized image (see FIG. 11D) (step C6). Here, in the binarization process, the image processing unit 16 extracts the luminance value (data (x, y)) of the pixel to be processed from the difference processed image (step D1), and the luminance value of the pixel and the threshold value (The threshold value corrected in step C3) is compared to determine whether the luminance value of the pixel is equal to or higher than the threshold value (step D2). If the luminance value of the pixel is equal to or higher than the threshold value (step D2) If YES, the pixel brightness value is 1 (white) (step D3). If the pixel brightness value is not greater than or equal to the threshold value (NO in step D2; no scratch area), the pixel brightness When the value is set to 0 (black) (step D4), it is confirmed whether or not the binarization processing has been performed for all the pixels of the difference processed image (step D5). When the binarization processing has not been performed for all the pixels (step D5) Return to step D1 for NO in step D5) Performs processing for the next pixel, the process proceeds to step C7 when performing binarization processing for all the pixels (YES in step D5).

  Next, the image processing unit 16 calculates feature quantities such as the length and area of each flawed portion from the binarized image (step C7).

  Finally, the image processing unit 16 determines a non-defective product or a defective product by comparing the feature amount with a threshold value (fixed value) (step C8). In the determination, if the feature amount is equal to or greater than the threshold value, the product is determined to be defective. In the binarized image of FIG. 11 (d), there are three features whose feature amount is equal to or greater than the threshold value, and it is determined that the product is defective.

  According to the second embodiment, even when the illumination is deteriorated, the blade is replaced, etc., the threshold value is adjusted by the statistical method of the imaging screen to obtain stable determination performance.

Example 1
Next, the drawings are compared while comparing the defect inspection processing when a fixed value is used as the threshold value used in the binarization processing (in the case of the first embodiment) and when correction values are used (in the case of the second embodiment). It explains using. 12A and 12B are images (images when the illumination brightness is 1) obtained by the defect inspection processing according to the comparative example, where FIG. 12A is an original image, FIG. 12B is a filtered image, and FIG. 12C is a difference processed image. , (D) are binarized images. FIG. 13 shows scratches related to defect inspection processing when a fixed value is used as a threshold value (in the case of the first embodiment) and when a correction value is used (in the case of the second embodiment) for each luminance of illumination. It is a table | surface for comparing the number of determination. The specimen 2 used here is a sample of three scratches.

  When the defect inspection process is performed without performing the dispersion value calculation process and the threshold value correction process in the image processing unit 16 when the illumination 15 has low luminance (for example, luminance 1) (that is, in the case of the first embodiment). An image as shown in FIG. 12 is obtained, and the processing mark is determined as a scratch. That is, when the threshold value used in the binarization process is a fixed value (in the case of the first embodiment), the processing mark is determined to be scratched when the brightness of the illumination 15 is 5 or less, but when the threshold value is corrected ( In the case of the second embodiment, an image as shown in FIG. 11 is obtained, and the illumination 15 can be accurately determined in a wide range from luminance 1 to 15. Even in the case of the first embodiment, if the brightness of the illumination 15 is 6 or more, the processing mark is not judged as a scratch, which is effective.

  For each luminance of illumination, the number of scratches determined when a fixed value is used as a threshold value (in the case of the first embodiment) and when a correction value is used (in the case of the second embodiment) is as shown in FIG. . When a fixed value is used, the number of processing marks to be judged as scratches increases with a luminance of 5 or less. However, when a correction value is used, three determinations can be made at any luminance.

(Example 2)
Next, in the case of performing defect inspection processing on a specimen that has been cut immediately after blade replacement, a fixed value is used (in the case of the first embodiment) and a correction value are used as threshold values used in the binarization processing. Description will be made using the drawings while comparing (in the case of Embodiment 2). FIG. 14 is an image for comparing the defect inspection process when using a sample having no scratch immediately after exchanging the cutting tool. FIG. 15 is an image for comparing defect inspection processing when a sample with scratches is used immediately after blade replacement. Here, the luminance of illumination is set to 15. This is because even if the brightness is further increased in order to erase unnecessary processing marks, the processing marks do not disappear, and three defects detected in FIG. 4 (high luminance) disappear due to illumination.

  A case where an image having only a processing mark (an image without a scratch) is used will be described. Immediately after replacing the cutting tool, the processing mark may not disappear even if the brightness of the illumination is increased depending on the shape of the cutting tool. When a fixed value is used as a threshold value used in the binarization processing for a non-defective image such as that shown in FIG. 14A, the case shown in FIG. ), The processing mark is judged as a scratch. On the other hand, when a correction value is used as a threshold value used in the binarization process (in the case of the second embodiment), even when a processing mark is generated due to the blade replacement, there is no scratch as shown in FIG. Can be determined.

  A case where an image in which processing marks and scratches are mixed will be described. When a fixed value is used as a threshold value used in the binarization processing for the original image in which processing marks and scratches are mixed as shown in FIG. As in 15 (b), it becomes impossible to determine the processing marks and scratches. On the other hand, when a correction value is used as the threshold value used in the binarization process (in the case of the second embodiment), only a scratch is detected as shown in FIG. Is possible.

  As described above, the threshold value is corrected based on the dispersion value (or average value) of the cutting surface as in the second embodiment, so that accurate determination can be made even when the measurement environment changes extremely, such as variations in illumination or replacement of cutting tools. In the first embodiment, patterns that cannot be accurately determined can be handled. Also, when determining or correcting the threshold value, in order to take into account the aging deterioration of the lighting, the time is measured from the start of use of the lighting, and the decrease in the illuminance of the LED or the heating lamp is used to determine the threshold value. It is also effective to reflect it.

It is the block diagram which showed typically the structure of the defect inspection apparatus which concerns on Embodiment 1 of this invention. It is the flowchart which showed typically the processing operation of the image process part of the defect inspection apparatus which concerns on Embodiment 1 of this invention. It is the flowchart which showed typically the binarization process of the image processing part of the defect inspection apparatus which concerns on Embodiment 1 of this invention. It is an image (image at the time of illumination brightness | luminance 15) obtained by the defect inspection apparatus which concerns on Embodiment 1 of this invention, (a) is an original image, (b) is a filter process image, (c) is a difference process image. , (D) are binarized images. It is the graph which showed typically the map (dispersion value-threshold value) used in the image processing part of the defect inspection apparatus which concerns on Embodiment 2 of this invention. It is a graph for demonstrating the relationship between the brightness | luminance of illumination, and a total value in the defect inspection apparatus which concerns on Embodiment 2 of this invention. It is a graph for demonstrating the relationship between the brightness | luminance of illumination, and a dispersion value in the defect inspection apparatus which concerns on Embodiment 2 of this invention. It is a graph for demonstrating the relationship between the brightness | luminance of illumination and a threshold value (correction value) in the defect inspection apparatus which concerns on Embodiment 2 of this invention. It is the flowchart which showed typically the processing operation of the image process part of the defect inspection apparatus which concerns on Embodiment 2 of this invention. It is the flowchart which showed typically the binarization process of the image processing part of the defect inspection apparatus which concerns on Embodiment 2 of this invention. It is an image (image at the time of illumination brightness | luminance 1) obtained by the defect inspection apparatus which concerns on Embodiment 2 of this invention, (a) is an original image, (b) is a filter process image, (c) is a difference process image. , (D) are binarized images. It is an image (image when the luminance of illumination is 1) obtained by the defect inspection apparatus according to the comparative example, (a) is the original image, (b) is the filtered image, (c) is the difference processed image, (d) Is a binarized image. In order to compare the determination number of scratches related to defect inspection processing when a fixed value is used as a threshold value for each luminance of illumination (in the case of Embodiment 1) and when a correction value is used (in the case of Embodiment 2) It is a table. It is an image for comparing defect inspection processing when using a sample without scratch immediately after exchanging the blade. It is an image for comparing defect inspection processing when using a scratched sample immediately after replacing the blade.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Defect inspection apparatus 2 Specimen 11 Turntable 12 Motor 13 Encoder 14 Camera 15 Illumination 16 Image processing part 17 Display part

Claims (8)

A defect inspection apparatus for inspecting a surface defect of a specimen,
A camera that images the surface of the specimen and acquires an original image;
A filter processing unit for creating a filtered image obtained by removing noise from the original image;
A threshold value determination unit for determining a threshold value based on information of pixels of the original image or pixels of the filtered image;
When the luminance value of the pixel of the filtered image is equal to or higher than the threshold value, the luminance value of the pixel is replaced with 1, and when the luminance value of the pixel of the filtered image is not equal to or higher than the threshold value, the luminance value of the pixel A binarization processing unit for creating a binarized image in which the value is replaced with 0;
A determination processing unit that determines the presence or absence of defects in the binarized image;
A defect inspection apparatus comprising: A difference processing unit that creates a difference processed image obtained by subtracting the luminance value of the corresponding pixel of the filtered image from the luminance value of the pixel of the original image;
The binarization processing unit replaces the luminance value of the pixel with 1 when the luminance value of the pixel of the difference processing image is equal to or greater than a threshold value, and the luminance value of the pixel of the difference processing image is equal to or greater than the threshold value. 2. The defect inspection apparatus according to claim 1, wherein a binary-processed image in which the luminance value of the pixel is replaced with 0 is created when the pixel is not.   The defect according to claim 1, wherein the threshold value determination unit calculates an average value from luminance values of pixels of the original image or pixels of the filtered image, and determines the threshold value. Inspection device.   3. The defect according to claim 1, wherein the threshold value determination unit calculates a variance value from a luminance value of a pixel of the original image or a pixel of the filtered image and determines the threshold value. Inspection device.   The filter processing unit calculates an average value of the luminance values of the target pixel and its surrounding pixels based on the original image, and creates a filtered image in which the calculated average value and the luminance value of the target pixel are switched. The defect inspection apparatus according to any one of claims 1 to 4.   The filter processing unit extracts a maximum value of luminance values of a target pixel and its surrounding pixels based on the original image, and creates a filtered image in which the extracted maximum value and the luminance value of the target pixel are switched. The defect inspection apparatus according to any one of claims 1 to 4. Illuminating means for illuminating the surface of the specimen,
The defect inspection apparatus according to claim 1, wherein the threshold value determination unit determines the threshold value based on aged deterioration of the illumination unit. A defect inspection method for inspecting a surface defect of a specimen,
Capturing the original image by imaging the surface of the specimen;
Creating a filtered image from which noise has been removed from the original image;
Determining a threshold based on information of pixels of the original image or pixels of the filtered image;
When the luminance value of the pixel of the filtered image is equal to or higher than the threshold value, the luminance value of the pixel is replaced with 1, and when the luminance value of the pixel of the filtered image is not equal to or higher than the threshold value, the luminance value of the pixel Creating a binarized image with the value replaced with 0;
Determining the presence or absence of defects in the binarized image;
A defect inspection method comprising:

Images