Defect detection method and device for image data of integrated circuit AVI equipment

Abstract

The invention discloses an integrated circuit AVI equipment image data-oriented defect detection method and device, and the method comprises the steps: obtaining initial complete images of a plurality of PCB support plates of the same model, carrying out the geometric registration, brightness normalization and image fusion processing of the initial complete images, and constructing a standard image set; obtaining an initial complete image A of the PCB carrier plate to be detected, and comprehensively comparing the initial complete image A with the standard image set to obtain a plurality of feature difference values of each pixel point in the initial complete image A; carrying out weighted average on the plurality of feature difference values of each pixel point to generate a difference chart of the PCB support plate to be detected; and performing defect judgment and marking on the difference chart of the to-be-detected PCB support plate based on a preset threshold value. The defect detection method is reasonable and easy to operate and implement, the accuracy, robustness and interpretability of defect detection work can be remarkably improved, and the defect detection method can be well applied to the manufacturing process of integrated circuits and PCB carrier plates.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic manufacturing, in particular to a defect detection method and device for integrated circuit AVI equipment image data. BACKGROUND

[0002] With the continuous improvement of the manufacturing process of integrated circuits and PCB carriers, and the increasingly complex device structure, higher requirements are put forward for the precision and efficiency of the product detection link.

[0003] Traditional AVI equipment usually acquires multiple local images of PCB products by using a camera, and then performs defect detection on the obtained images. However, on the one hand, since multiple local images of the PCB product are detected and analyzed, the product defect omission rate is high, thereby causing low detection accuracy. On the other hand, since the existing defect detection method mainly compares based on a single feature or a single color space, it is easily affected by light, color difference and image noise, thereby causing high misjudgment and omission rates, further reducing the detection accuracy.

[0004] Therefore, how to improve the accuracy and robustness of defect detection has become a technical problem to be solved. In view of this, the present application is proposed. SUMMARY

[0005] In order to overcome the above defects, the present application provides a defect detection method and device for integrated circuit AVI equipment image data. The defect detection method is reasonable and easy to operate and implement, can significantly improve the accuracy, robustness and explainability of defect detection work, and can be well applied to the manufacturing process of integrated circuits and PCB carriers.

[0006] The technical scheme adopted by the present application to solve its technical problems is: a defect detection method for integrated circuit AVI equipment image data, comprising: Obtaining initial complete images of multiple PCB carriers of the same model, and sequentially performing geometric registration, brightness normalization and image fusion processing on the obtained initial complete images of the PCB carriers to construct a standard image set; Obtaining an initial complete image A of a to-be-detected PCB carrier, and comprehensively comparing the initial complete image A with the standard image set to obtain multiple feature difference values of each pixel point in the initial complete image A; Weighted average is performed on the multiple feature difference values of each pixel point in the initial complete image A to generate a difference map of the to-be-detected PCB carrier; and then defect judgment and labeling are performed on the obtained difference map of the to-be-detected PCB carrier based on a preset threshold.

[0007] As a further improvement of the present application, the method for obtaining the initial complete image of each PCB carrier plate is: first, using a line scanning camera to obtain multiple local images of the PCB carrier plate, and then using an image feature matching algorithm to stitch the multiple local images into an initial complete image.

[0008] As a further improvement of the present application, the initial complete images of M pieces of PCB carrier plates are obtained, and the initial complete image of the kth PCB carrier plate is defined as , wherein ; The method for performing geometric registration on the obtained initial complete image of the PCB carrier plate is: first, calculating the transformation matrix of the initial complete image of the kth PCB carrier plate using an image feature matching algorithm, and then mapping the initial complete image to a unified reference coordinate system according to the inverse matrix of the obtained transformation matrix to obtain a registered image ; that is, the registered image satisfies the following relationship with the initial complete image : ; Performing brightness normalization processing on the obtained registered image obtains the standard complete images of the multiple pieces of PCB carrier plates; Performing image fusion on the obtained standard complete images of the multiple pieces of PCB carrier plates to construct the standard image set, wherein the specific image fusion method is: defining the standard complete image of the kth PCB carrier plate as , collecting the standard complete images of the multiple pieces of PCB carrier plates to generate a standard image set , and simultaneously using the median method to optimize the obtained standard image set to eliminate abnormal values therein.

[0009] As a further improvement of the present application, RGB comparison, gradient comparison and HSV comparison are performed between the initial complete image A and the standard image set to obtain the RGB feature difference value, the gradient feature difference value and the HSV feature difference value of each pixel point in the initial complete image A.

[0010] As a further improvement of the present application, the method for performing RGB comparison between the initial complete image A and the standard image set is: calculating the absolute value of the difference value of the corresponding pixel points of the initial complete image A and the standard image set in the RGB three channels, and taking the maximum value as the RGB feature difference value of the pixel point in the initial complete image A. The method for gradient comparison between the initial complete image A and the standard image set is: first, the horizontal gradient and the vertical gradient of the corresponding pixel points of the initial complete image A and the standard image set are calculated respectively, then the square sum of the gradient difference of the corresponding pixel points of the initial complete image A and the standard image set is calculated, and the obtained square sum of the gradient difference is taken as the gradient feature difference value of the pixel point in the initial complete image A; The method for HSV comparison between the initial complete image A and the standard image set is: first, the initial complete image A and the standard image set are converted from RGB space to HSV space respectively, then the absolute value of the difference value of the corresponding pixel points of the initial complete image A and the standard image set on the three channels of HSV is calculated, and the maximum value among them is taken as the HSV feature difference value of the pixel point in the initial complete image A.

[0011] As a further improvement of the present application, the RGB feature difference value, the gradient feature difference value and the HSV feature difference value of each pixel point in the initial complete image A are weighted and averaged to obtain the comprehensive difference value of each pixel point, and then the difference map of the to-be-inspected PCB carrier is generated according to the comprehensive difference value of each pixel point in the initial complete image A.

[0012] As a further improvement of the present application, the difference map of the to-be-inspected PCB carrier is compared with a preset threshold, if: the comprehensive difference value of a certain pixel point in the difference map is greater than the preset threshold, it is determined that the pixel point has a defect, otherwise, it is determined that the pixel point is normal.

[0013] As a further improvement of the present application, the difference map of the to-be-inspected PCB carrier is textually labeled, and the labeled content includes: there is a defect anomaly at a certain position of the product, or there is no defect anomaly of the product.

[0014] The present application also provides a defect detection device for integrated circuit AVI equipment image data, comprising a processor, which can realize the steps of the defect detection method for integrated circuit AVI equipment image data.

[0015] The application has the following advantages: ①The application can provide more comprehensive product information by stitching multiple local images of a PCB carrier into a complete large image that can reflect the overall appearance of the PCB carrier, which is more convenient to find possible defects and problems on the PCB carrier, thus positively promoting and ensuring the comprehensiveness, accuracy and robustness of the comprehensive comparison and analysis operation of the subsequent process. ②The application can effectively reduce random errors in batch detection by introducing the standard image set as a comparison reference, which significantly improves the stability of the defect detection result. ③The application performs RGB comparison, gradient comparison and HSV comparison between the initial complete image A of the PCB carrier to be detected and the standard image set, which can capture color deviation, structural denaturation and abnormal illumination in the image at the same time, thus greatly reducing the misjudgment rate and the omission rate, and significantly improving the accuracy, robustness and interpretability of image matching and abnormal detection operation. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The flowchart of the defect detection method for integrated circuit AVI equipment image data in embodiment 1 of the application. DETAILED DESCRIPTION

[0017] The preferred embodiments of the application will be described in detail below with reference to the accompanying drawings.

[0018] Embodiment 1

[0019] Please refer to the accompanying drawings Figure 1 The embodiment 1 provides a defect detection method for integrated circuit AVI equipment image data, which includes the following detection steps: S1: Obtain the initial complete image of multiple PCB carriers of the same model, and perform geometric registration, brightness normalization and image fusion processing on the obtained initial complete image of the PCB carrier to construct a standard image set (Golden Image).

[0020] Specifically, according to the production requirement, the embodiment obtains the initial complete image of M PCB carriers, and the preferred method for obtaining the initial complete image of each PCB carrier is as follows: first, use a line scanning camera to obtain multiple local images of each PCB carrier, and then stitch the multiple local images into an initial complete image through an image feature matching algorithm.

[0021] Understandably, ① the line-scan camera acquires image information through line-by-line scanning, so the embodiment can ensure high image resolution and imaging quality when using the line-scan camera to collect image information of a PCB carrier with a certain length and width, and is particularly suitable for PCB carrier detection scenarios with high detail requirements. ② The image feature matching algorithm determines the relative positional relationship between images by finding common feature points such as corner points, edges, and textures between different images. Specifically, the image feature matching algorithm analyzes the features in each local image of each PCB carrier, then finds matching feature pairs between different local images, and calculates transformation parameters such as translation, rotation, and scaling based on these matching feature pairs. Through these transformation parameters, multiple local images are spatially aligned and spliced to obtain a complete large image that can reflect the overall appearance of the entire PCB carrier, i.e., an initial complete image of the PCB carrier. Understandably, obtaining the initial complete image of the PCB carrier is crucial for subsequent comprehensive comparison and analysis, because the complete large image can provide more comprehensive information and is more convenient for discovering possible defects and problems on the PCB carrier.

[0022] Further, regarding the above-mentioned image feature matching algorithm, the present application adopts but is not limited to the traditional image feature matching algorithm, which includes the following specific steps: feature detection (which can be implemented by using the scale-invariant feature transform algorithm SIFT), feature description (which can be implemented by using the SIFT descriptor), feature matching (which can be implemented by using the brute-force matching algorithm or the fast library for approximate nearest neighbors matching algorithm FLANN), and matching screening (which can be implemented by using the nearest neighbor distance ratio algorithm NNDR or the random sample consensus algorithm RANSAC). Since it belongs to conventional technical means, it is not described in detail here.

[0023] Based on the obtained initial complete images of the M pieces of PCB carriers, the embodiment sequentially performs geometric registration, brightness normalization and image fusion processing on them to construct the standard image set (Golden Image). It can be understood that the embodiment adopts the image processing method of combining geometric registration, brightness normalization and image fusion to construct the standard image set, because: ① in the actual image acquisition process, there may be differences in position, angle and scale between the obtained initial complete images of the M pieces of PCB carriers. In order to accurately fuse these initial complete images, it is necessary to map them to a unified reference coordinate system. That is, the embodiment can eliminate the geometric differences between different initial complete images by performing geometric registration on the initial complete images of the M pieces of PCB carriers, so that the subsequent fusion operation can be performed in the same coordinate system, ensuring the accuracy and consistency of the fusion result. ② In the actual image acquisition process, the images acquired in different batches may be affected by changes in lighting conditions, resulting in brightness differences between the images. Such brightness differences will interfere with subsequent image fusion and comparison, affecting the accuracy of the detection result. Therefore, the embodiment can eliminate the effects of batch illumination and color difference by performing brightness normalization on the initial complete images of the M pieces of PCB carriers, so that all initial complete images have consistency in brightness. In summary, the embodiment performs geometric registration and brightness normalization on the initial complete images of the PCB carriers, and then performs image fusion, which lays a good foundation for constructing the standard image set with high accuracy and consistency.

[0024] Further, the embodiment performs the geometric registration method on the initial complete images of the plurality of PCB carriers as follows: first, define the initial complete images of the plurality of PCB carriers as follows: the initial complete image of the kth PCB carrier is (i.e. the pixel value at the coordinate ), and ; it can be understood that the initial complete image of the first PCB carrier is , the initial complete image of the second PCB carrier is , and the initial complete image of the Mth PCB carrier is ; then, calculate the transformation matrix of each initial complete image of the PCB carrier by an image feature matching algorithm, such as the transformation matrix of the initial complete image of the kth PCB carrier is The process of calculating the transformation matrix of the initial complete image by the image feature matching algorithm is similar to the above-mentioned process of image stitching by the image feature matching algorithm, specifically: the feature points in the initial complete image can be detected and matched by SIFT algorithm or ORB algorithm, etc., based on the matched point pairs, and the RANSAC algorithm is used to calculate the transformation matrix Then, the initial complete image is mapped to the unified reference coordinate system according to the inverse matrix of the obtained transformation matrix to obtain the registered image It can be understood that the registered image satisfies the following relationship with the initial complete image .

[0025] Based on the obtained registered image , the brightness normalization processing is performed to obtain the standard complete images of the multiple PCB carriers. The specific algorithm of brightness normalization processing can be, but is not limited to, histogram equalization, gray world algorithm, etc., and is specifically determined according to the characteristics of the image and actual needs, as long as different images can be as close as possible in brightness.

[0026] Based on the obtained standard complete images of the multiple PCB carriers, the image fusion is performed to construct the standard image set. The implementation method of the image fusion of the standard complete images of the multiple PCB carriers in the embodiment is as follows: first, the standard complete images of the multiple PCB carriers are defined as follows: the standard complete image of the kth PCB carrier is , and The description of the standard complete image of each PCB carrier can be referred to the above description of the initial complete image of each PCB carrier, and thus is not repeated here; then, the standard complete images of the multiple PCB carriers are collected to generate the standard image set , and the obtained standard image set is optimized by using the median method, i.e. , to eliminate the outliers in the obtained standard image set . It can be understood that the use of the median method can effectively reduce the influence of abnormal factors in individual images on the obtained standard image set, so that the generated standard image set is more accurate, stable and reliable, and can more accurately reflect the ideal state of the same type of PCB carrier.

[0027] ​S2: Obtain an initial complete image A of the to-be-inspected PCB carrier board, and comprehensively compare the initial complete image A with the standard image set to obtain a plurality of feature difference values of each pixel point in the initial complete image A.

[0028] Specifically, the method for obtaining the initial complete image A of the to-be-inspected PCB carrier board in the embodiment can refer to the method for obtaining the initial complete image of each PCB carrier board in S1 described above, and thus will not be repeated here.

[0029] In the embodiment, RGB comparison, gradient comparison and HSV comparison are performed between the initial complete image A and the standard image set to correspondingly obtain the RGB feature difference value, the gradient feature difference value and the HSV feature difference value of each pixel point in the initial complete image A. It can be understood that, by performing the above three-modal feature comparison between the initial complete image A and the standard image set, the color deviation (RGB comparison), the structural denaturation (gradient comparison) and the illumination abnormality (HSV comparison) in the image can be captured at the same time, so that the accuracy, robustness and interpretability of image matching and abnormality detection can be significantly improved.

[0030] Further, the preferred method for performing RGB comparison between the initial complete image A and the standard image set in the embodiment is to calculate the absolute values of the difference values of the corresponding pixel points of the initial complete image A and the standard image set in the RGB three channels, and take the maximum value as the RGB feature difference value of the pixel point in the initial complete image A. Specifically, if the RGB channel value of a certain pixel point in the initial complete image A is defined as , the RGB channel value of the pixel point at the corresponding position in the standard image set is , the absolute values of the difference values of the corresponding pixel points of the initial complete image A and the standard image set in the RGB three channels are calculated as , and , and the RGB feature difference value of the pixel point is .

[0031] The preferred method for performing gradient comparison between the initial complete image A and the standard image set in the embodiment is to first calculate the horizontal gradient and the vertical gradient of the corresponding pixel points of the initial complete image A and the standard image set, and then calculate the sum of squares of the gradient difference of the corresponding pixel points of the initial complete image A and the standard image set, and take the sum of squares of the gradient difference as the gradient feature difference value of the pixel point in the initial complete image A. Specifically, in the embodiment, the horizontal gradient and the vertical gradient of the pixel point use the first-order difference of the pixel value, that is, the horizontal gradient longitudinal gradient If the gradient of a certain pixel in the initial complete image A is defined as... The gradient of the pixel at the corresponding position in the standard image set is Correspondingly, the gradient feature difference value of this pixel is: .

[0032] The preferred method for HSV comparison between the initial complete image A and the standard image set in this embodiment is as follows: First, convert the initial complete image A and the standard image set from RGB space to HSV space respectively. Then, calculate the absolute values ​​of the differences between corresponding pixels in the initial complete image A and the standard image set across the three HSV channels, and take the maximum value as the HSV feature difference value of that pixel in the initial complete image A. Specifically, if the HSV channel value of a certain pixel in the initial complete image A is defined as... The HSV channel value of the pixel at the corresponding position in the standard image set is The absolute values ​​of the differences between the corresponding pixels of the initial complete image A and the standard image set in the three HSV channels are calculated as follows: , and At that time, the HSV feature difference value of that pixel will be: .

[0033] S3: Perform a weighted average of multiple feature difference values ​​for each pixel in the initial complete image A to generate a difference map of the PCB carrier to be inspected; then, based on a preset threshold, determine and label the defects in the obtained difference map of the PCB carrier to be inspected.

[0034] Specifically, in this embodiment, the RGB feature difference value, gradient feature difference value, and HSV feature difference value of each pixel in the initial complete image A are weighted and averaged to obtain the comprehensive difference value of each pixel in the initial complete image A; that is, the comprehensive difference value of each pixel in the initial complete image A. The calculation formula is: ,in, , and The weighting coefficients are all not less than 0 and simultaneously satisfy the following conditions: Then, a difference map of the PCB carrier board to be inspected is generated based on the comprehensive difference value of each pixel in the initial complete image A. Next, the difference map of the PCB carrier board to be inspected is compared with a preset threshold, and the comparison logic is as follows: if the comprehensive difference value of a certain pixel in the difference map is greater than the preset threshold, then the pixel is determined to be defective; otherwise, the pixel is determined to be normal.

[0035] Further, the embodiment also performs text annotation on the difference map of the to-be-inspected PCB carrier board, and the annotated content includes: a certain position of the product has a defect anomaly, or the product has no defect anomaly.

[0036] As can be seen from the above, the defect detection method provided by the embodiment 1 has the following advantages: ①The embodiment 1 splices multiple local images of a PCB carrier board into one complete large image (i.e., the initial complete image of the PCB carrier board) that can reflect the overall appearance of the PCB carrier board by using an image feature matching algorithm, which can provide more comprehensive product information and facilitate the discovery of possible defects and problems on the PCB carrier board, thus positively promoting and guaranteeing the comprehensiveness, accuracy and robustness of the comprehensive comparison and analysis operation of the post-process. ②The embodiment 1 introduces the standard image set as a comparison reference, which can effectively reduce the random error in batch detection and significantly improve the stability of the defect detection result. ③The embodiment 1 performs RGB comparison, gradient comparison and HSV comparison between the initial complete image A of the to-be-inspected PCB carrier board and the standard image set, which can simultaneously capture color deviation, structural denaturation and illumination anomaly in the image, thereby greatly reducing the misjudgment rate and the omission rate and significantly improving the accuracy, robustness and interpretability of the image matching and anomaly detection operation.

[0037] Embodiment 2 The embodiment 2 provides a defect detection device for integrated circuit AVI equipment image data, which comprises a processor, and the processor can realize the steps of the defect detection method for integrated circuit AVI equipment image data in the above-mentioned embodiment 1.

[0038] Specifically, the processor can include one or more processing cores. The processor connects various parts in the entire device through various interfaces and lines, executes various functions and processes data of the device by running or executing instructions, programs, code sets or instruction sets stored in the memory, and calling data stored in the memory.

[0039] Further, the processor can be implemented in at least one of the hardware forms of a digital signal processing DSP, a field programmable gate array FPGA and a programmable logic array PLA. Alternatively, the processor can be integrated with one or a combination of a central processing unit CPU and a modem.

[0040] In addition, since the defect detection device provided by the embodiment 2 corresponds to the defect detection method provided by the above-mentioned embodiment 1, and the principle of solving problems of the defect detection device is similar to that of the defect detection method, the implementation of the defect detection device can be referred to the implementation process in the above-mentioned embodiment 1, and thus is not described herein.

[0041] Finally, the suffixes "A" and the like of the component names in the present application specification (such as the initial full image A and the like) are merely for the convenience of clear description, and are not intended to limit the scope of the present application.

[0042] In the above description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the above description is merely the preferred embodiment of the present application, and the present application can be implemented in many different ways from those described herein, and therefore the present application is not limited by the specific implementations disclosed above. Meanwhile, any person skilled in the art can make many possible changes and modifications to the technical solutions of the present application, or modify equivalent embodiments, by using the methods and technical contents disclosed above, without departing from the scope of the technical solutions of the present application. Any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application, without departing from the content of the technical solutions of the present application, shall still fall within the scope of protection of the technical solutions of the present application.

Claims

1. A defect detection method for image data of integrated circuit AVI devices, characterized in that: include: The initial complete images of multiple PCB carrier boards of the same model are obtained, and the obtained initial complete images of the PCB carrier boards are sequentially processed by geometric registration, brightness normalization and image fusion to construct a standard image set. An initial complete image A of the PCB carrier board to be inspected is obtained, and the initial complete image A is compared with the standard image set to obtain multiple feature difference values ​​for each pixel in the initial complete image A. A weighted average of multiple feature difference values ​​for each pixel in the initial complete image A is performed to generate a difference map of the PCB carrier to be inspected; then, based on a preset threshold, defects are determined and annotated in the obtained difference map of the PCB carrier to be inspected.

2. The defect detection method for integrated circuit AVI device image data according to claim 1, characterized in that: The method for obtaining the initial complete image of each PCB carrier is as follows: first, multiple local images of the PCB carrier are obtained using a line scan camera, and then the multiple local images are stitched together into an initial complete image using an image feature matching algorithm.

3. The defect detection method for integrated circuit AVI device image data according to claim 1, characterized in that: Obtain the initial complete image of M PCB carrier boards, and define the initial complete image of the k-th PCB carrier board as... ,in ; The method for geometric registration of the obtained initial complete image of the PCB carrier is as follows: First, the initial complete image of the k-th PCB carrier is calculated using an image feature matching algorithm. Transformation matrix Then, based on the obtained transformation matrix inverse matrix The initial complete image Mapping to a unified reference coordinate system yields the registered image. That is, the registered image With the initial complete image The following relationship exists between them: ; The resulting registered image Brightness normalization processing is performed to obtain standard complete images of multiple PCB carrier boards; The obtained standard complete images of multiple PCB carrier boards are fused to construct the standard image set. Specifically, the image fusion method is as follows: the standard complete image of the k-th PCB carrier board is defined as... Collect standard complete images of multiple PCB carrier boards to generate a standard image set. Meanwhile, the median method was used to analyze the obtained standard image set. Optimize to remove outliers.

4. The defect detection method for integrated circuit AVI device image data according to claim 1, characterized in that: RGB comparison, gradient comparison, and HSV comparison are performed between the initial complete image A and the standard image set to obtain the RGB feature difference value, gradient feature difference value, and HSV feature difference value of each pixel in the initial complete image A.

5. The defect detection method for integrated circuit AVI device image data according to claim 4, characterized in that: The method for performing RGB comparison between the initial complete image A and the standard image set is as follows: calculate the absolute value of the difference between the corresponding pixel points of the initial complete image A and the standard image set in the three RGB channels, and take the maximum value as the RGB feature difference value of the pixel point in the initial complete image A; The method for performing gradient comparison between the initial complete image A and the standard image set is as follows: First, calculate the horizontal gradient and vertical gradient of the corresponding pixels of the initial complete image A and the standard image set respectively. Then, calculate the sum of squares of the gradient differences of the corresponding pixels of the initial complete image A and the standard image set, and use the sum of squares of the obtained gradient differences as the gradient feature difference value of the pixel in the initial complete image A. The method for comparing the initial complete image A with the standard image set is as follows: First, convert the initial complete image A and the standard image set from RGB space to HSV space respectively. Then, calculate the absolute value of the difference between the corresponding pixel points of the initial complete image A and the standard image set in the three HSV channels, and take the maximum value as the HSV feature difference value of the pixel point in the initial complete image A.

6. The defect detection method for integrated circuit AVI device image data according to claim 5, characterized in that: The RGB feature difference value, gradient feature difference value, and HSV feature difference value of each pixel in the initial complete image A are weighted and averaged to obtain the comprehensive difference value of each pixel; then, the difference map of the PCB carrier to be inspected is generated based on the comprehensive difference value of each pixel in the initial complete image A.

7. The defect detection method for integrated circuit AVI device image data according to claim 6, characterized in that: The difference map of the PCB carrier board to be inspected is compared with a preset threshold. If the overall difference value of a certain pixel in the difference map is greater than the preset threshold, then the pixel is determined to have a defect. Conversely, if the pixel does not meet the requirements, it is considered normal.

8. The defect detection method for integrated circuit AVI device image data according to claim 7, characterized in that: Text annotations are added to the difference diagram of the PCB carrier board to be inspected, and the annotations include: there is a defect or abnormality at a certain location of the product, or there is no defect or abnormality in the product.

9. A defect detection device for image data of integrated circuit AVI devices, characterized in that: The device includes a processor capable of implementing the steps of the defect detection method for image data of integrated circuit AVI devices as described in any one of claims 1-8.

Images