A CMOS chip surface defect detection method and system

Abstract

The present invention discloses a method and system for detecting surface defects of CMOS chips, wherein the detection method includes the following steps: dividing the image area in the image data of the product to be tested to form multiple detection blocks; binarizing each of the detection blocks according to the grayscale value; extracting defects in each of the detection blocks according to the results of the binarization; and performing pixel and / or size analysis on the defects to further extract effective defects. The method of the present invention can accurately, quickly, and stably detect dirt defects on the surface of the product before the CMOS chip is filmed. The requirements for the grayscale consistency and uniformity of the overall image are reduced, the installation and debugging of the hardware light source consistency are facilitated, the consistency of the grayscale of local areas in the product is ensured, and it is compatible with the overall grayscale changes caused by material and processing differences of different products. The detection accuracy of defects with small grayscale contrast differences and subtle defects is higher.

Description

Technical Field

[0001] The present invention belongs to the field of semiconductor detection, and in particular relates to a CMOS chip surface defect detection method and system. Background Art

[0002] Before the lamination process in CMOS production, in order to ensure that the products loaded onto the carrier are free of defects, a surface inspection of the entire CMOS is required to eliminate products with appearance defects caused by mechanical collisions or unqualified workmanship.

[0003] Existing technologies generally rely on manual elimination, which is often slow and subjective. Furthermore, conventional techniques for detecting CMOS surface defects primarily rely on the grayscale difference between the defect and the normal surface to detect defects. This has the following drawbacks: a) it requires high grayscale consistency across the image, resulting in a narrow tolerance between different products; b) it only detects defects with significant grayscale contrast differences; and c) it lacks high detection accuracy for subtle surface defects. Summary of the Invention

[0004] In order to solve the above problems, the present invention adopts a technical solution: a method for detecting surface defects of a CMOS chip, the method comprising the following steps:

[0005] Divide the image area in the image data of the product to be tested into multiple detection blocks;

[0006] Performing binarization processing on each detection block according to the grayscale value;

[0007] Extracting defects in each of the detection blocks according to the result of the binarization process;

[0008] Perform pixel and / or size analysis on the defects to further extract valid defects.

[0009] Optionally, before the step of dividing the image area, the detection method further comprises the following steps:

[0010] Performing binarization processing on the image data of the product to be tested according to the grayscale value;

[0011] The image data of the product to be tested is subjected to edge detection processing on all four sides according to the binarization processing result, and a detection area is divided according to the edge detection result of all four sides. The detection area division result includes the image area and the non-image area.

[0012] Optionally, the detection method further comprises the steps of:

[0013] Extracting the image area and obtaining the positioning center of the image data of the product to be tested;

[0014] Calculating the position deviation of the image data of the product to be measured according to the positioning center;

[0015] A positioning offset is performed according to the position deviation to redefine the image area and the non-image area.

[0016] Optionally, in the step of performing binarization processing on each detection block according to the grayscale value, specifically:

[0017] Obtaining the grayscale median value of each detection block;

[0018] Each detection block is binarized according to upper and lower set differences compared to the grayscale median value.

[0019] Optionally, the detection method further comprises the steps of:

[0020] Performing secondary binarization processing on the extracted defect and the area surrounding the defect according to the grayscale value;

[0021] Defects are re-extracted according to the result of the secondary binarization process.

[0022] Optionally, in the step of performing pixel and size analysis on the defect, specifically:

[0023] Obtain the circumscribed rectangle and area of ​​the defect outline;

[0024] Determining whether the pixels of the circumscribed rectangle and the defect outline meet the requirements of a valid defect;

[0025] If yes, it is determined whether the size of the circumscribed rectangle and the area meets the card control standard for defective products.

[0026] Optionally, a secondary analysis is performed on the defects within a certain value close to the card control standard, and the secondary analysis is specifically as follows:

[0027] Obtaining the maximum grayscale value, the minimum grayscale value, and the average grayscale value of the defect;

[0028] Compare the maximum grayscale value, the minimum grayscale value, the grayscale mean value of the defect and the grayscale median value of the detection block where the defect is located with a set fixed value;

[0029] Re-extract the defects in the detection block according to the comparison result, and exclude the edge transition pixels of the defects;

[0030] The bounding rectangle and area of ​​the defect are retrieved.

[0031] Optionally, the detection method further comprises the following steps:

[0032] Selecting a fixed grayscale value for a non-image area of ​​the image data of the product to be tested;

[0033] performing a binarization process on each of the non-image areas according to the comparison with the fixed grayscale value;

[0034] Defects in the non-image area are extracted according to the result of the binarization process.

[0035] Optionally, before performing binarization processing on the non-image area, the method further includes the following steps:

[0036] Eliminate the influence of edge lines in the non-image area of ​​the product to be tested.

[0037] And, a CMOS chip surface defect detection system, comprising:

[0038] Image acquisition module, used to collect image data of the product to be tested;

[0039] A block division module is used to divide the image area in the image data of the product to be tested and form a plurality of detection blocks for defect detection;

[0040] The defect extraction module includes: a binarization processing unit for performing binarization processing on each of the detection blocks according to the grayscale value; a defect extraction unit for extracting defects in the plurality of detection blocks;

[0041] The defect analysis module is used to perform pixel and / or size analysis on the extracted defects to further extract effective defects.

[0042] Optionally, the binarization processing unit is further configured to perform binarization processing on the image data of the product to be tested according to grayscale values;

[0043] The detection system also includes a region division module, which performs edge detection processing on the image data of the product to be tested based on the binarization processing result, and divides the detection area according to the edge detection result of the four edges. The detection area division result includes the image area and the non-image area.

[0044] Optionally, the detection system further includes a positioning module, which is used to calculate the position deviation of the image data of the product to be tested and re-determine the image area and non-image area of ​​the image data of the product to be tested.

[0045] Optionally, the binarization processing unit is further used to perform binarization processing on the non-image area of ​​the image data of the product to be tested according to the grayscale value, and the defect extraction unit is further used to extract defects in the non-image area according to the processing result of the binarization processing unit.

[0046] Optionally, the binarization processing unit is further configured to perform secondary binarization processing on the extracted defect and the defect surrounding area according to grayscale values, and the defect extraction unit is further configured to re-extract defects from the defect and the defect surrounding area after the secondary binarization processing.

[0047] Optionally, the defect analysis module includes:

[0048] An analysis unit, configured to obtain a circumscribed rectangle and an area of ​​the defect outline, and determine whether the pixels and size of the circumscribed rectangle and the size of the area meet the valid defect and defective product control standards;

[0049] The processing unit is used to analyze and process the defects within a certain value close to the card control standard, and redetermine the circumscribed rectangle and area of ​​the defects after eliminating the influence of the edge transition pixels of the defects.

[0050] The present invention adopts the above technical solution, so it has the following beneficial effects:

[0051] 1. It can accurately, quickly and stably detect dirt defects on the product surface before CMOS chip film is attached;

[0052] 2. The requirements for the grayscale consistency and uniformity of the entire image are reduced, which facilitates the installation and debugging of the hardware light source consistency and ensures the grayscale consistency of local areas in the product;

[0053] 3. It is compatible with the overall grayscale changes caused by material and processing differences of different products, and has higher detection accuracy for defects with small grayscale contrast differences and subtle defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0055] Figure 1 A flowchart of a method for detecting surface defects of a CMOS chip according to an embodiment of the present invention is shown;

[0056] Figure 2 shows a structural diagram of a CMOS chip after the first binarization process according to an embodiment of the present invention;

[0057] Figure 3 It shows a structural diagram of a CMOS chip according to an embodiment of the present invention after being divided into multiple detection blocks;

[0058] Figure 4 A schematic diagram showing the surface grayscale and defects of a CMOS chip;

[0059] Figure 5 FIG. 4 shows a structural diagram of edge line detection in a CMOS chip according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0061] In today's society, with the emergence and development of science and technology such as computer technology and artificial intelligence, as well as in-depth research, surface defect detection technology based on machine vision has emerged. The emergence of this technology has greatly improved production efficiency, avoided the influence of operating conditions and subjective judgment on the accuracy of detection results, and achieved better and more accurate surface defect detection and faster identification of product surface defects.

[0062] Product surface defect detection is a type of machine vision technology that uses computer vision to simulate human vision, capturing, processing, and calculating images from physical objects, ultimately leading to actual inspection, control, and application. Product surface defect detection is a crucial component of machine vision inspection, and its accuracy directly impacts the final product quality. Manual inspection methods have long been unable to meet the demands of production and modern manufacturing processes, and machine vision inspection has effectively overcome this. The widespread use of surface defect detection systems has promoted high-quality factory production and the development of intelligent automation in the manufacturing industry.

[0063] like Figure 1 The CMOS chip surface defect detection method according to an embodiment of the present invention comprises the following steps:

[0064] S1. Accurate positioning

[0065] The image data of the product to be tested is subjected to a first binarization process according to the grayscale value, such as Figure 2 The structure diagram of the CMOS chip after the first binarization process is shown;

[0066] Performing edge detection processing on the image data of the product to be tested based on the binarization processing result, and dividing the image area and the non-image area based on the edge detection result;

[0067] Extracting the image area and obtaining the positioning center of the image data of the product to be tested;

[0068] Calculating the position deviation of the image data of the product to be measured according to the positioning center;

[0069] A positioning offset is performed according to the position deviation to redetermine the positioning center.

[0070] Specifically, in this embodiment, a total of eight edge-finding parameters are set for both the image and non-image areas, including the region, edge-finding direction (inward or outward), grayscale transition direction (increasing or decreasing grayscale), grayscale transition, edge-finding interval, and edge point selection order. Generally, the image area is set to search outward for the first grayscale-decreasing transition point. A grayscale transition value greater than 20 indicates an edge line. The non-image area searches inward for the first grayscale-decreasing transition point. A grayscale transition value greater than 20 indicates an edge line. In this embodiment, a grayscale value of 255 corresponds to pure white, 0 corresponds to pure black, and 255→0 corresponds to the image transition from white to black.

[0071] S2. Regional Division

[0072] Since different parameters are used to extract defects in the image area and the non-image area, it is necessary to redefine the image area and the non-image area. Specifically, according to the redetermined positioning center, the image data of the product to be tested is binarized for the second time, and the image data of the product to be tested is edge-finding processed based on the result of the second binarization processing. The image area and the non-image area are divided according to the edge-finding results of the surrounding edges.

[0073] After the area division, the image area in the image data of the product to be tested is divided into multiple detection blocks, and the multiple detection blocks are independently extracted for defects. Figure 3 The structure of the CMOS chip shown is after being divided into multiple detection blocks, which are divided into 25 blocks with 5 rows and 5 columns.

[0074] S3. Defect extraction

[0075] 1) The defect extraction step for the image area is: performing binarization processing on each of the detection blocks according to the grayscale value, and extracting defects in each of the detection blocks according to the result of the binarization processing.

[0076] Specifically, the grayscale values ​​of the detection blocks are statistically analyzed through a histogram to obtain the grayscale median of each detection block, which is used as the adaptive threshold. Each detection block is binarized based on the upper and lower set differences compared to the grayscale median. Specifically, the adaptive threshold + the relative bright threshold are used for binarization to extract white spots, and the adaptive threshold - the relative dark threshold is used for binarization to extract black spots.

[0077] like Figure 4 The schematic diagram of the surface grayscale and defect grayscale of the CMOS chip shown in the figure shows that there is a grayscale value deviation between the upper and lower parts of the product. Since there are also overall grayscale differences between different products, in this solution, an adaptive algorithm is used to count the grayscale characteristics of the current product, which can avoid the differences between different products and facilitate the unification of detection parameters.

[0078] 2) The defect extraction step for the non-image area is as follows: performing binarization processing on the non-image area according to the grayscale value; and extracting defects in the non-image area according to the result of the binarization processing.

[0079] Specifically, the method involves selecting a fixed grayscale value for the non-image area of ​​the image data of the product to be tested; performing a binarization process on each non-image area based on the comparison with the fixed grayscale value; and extracting defects from the non-image area based on the binarization results. Because surface defects in non-image areas have a less significant impact on the product and require less stringent analysis, unlike image areas, the binarization process uses a set fixed grayscale value, rather than a statistical grayscale median. The range of fixed grayscale values ​​can be adjusted as needed.

[0080] In addition, before extracting defects in the non-image area, it is also necessary to eliminate the influence of the edge lines in the non-image area of ​​the product to be tested. There are many lines and features inside the non-image area, and they all need to set separate inspection parameters, including position area, inspection threshold, inspection value range, etc. The line area is fixed relative to the product, and the position does not change much. During training, its position relative to the image center, line direction, and size are recorded. During inspection, the corresponding inspection area is moved according to the change of the image center, and a separate binary analysis is performed on the inspection area to obtain the inspection value.

[0081] like Figure 5The structural diagram of edge line detection in the CMOS chip shown in the figure shows that the line between the two crosses at the bottom is the sealing line. Before extracting defects, the influence of the sealing line needs to be eliminated to avoid extracting the sealing line as a defect. The four corners of the image area are determined through the above-mentioned edge finding steps. The intersection of the corner connection line (the middle dotted line in the figure) is the image center (the middle cross position). The deviation in the XY direction from the two ends of the sealing line to the image center, as well as the upper and lower widths of the sealing line, are recorded. A grayscale value is used to detect defects in the area around the sealing line. When inspecting different products, there is a position deviation in the center cross determined by the product edge finding, but the deviation of the two ends of the line relative to the center cross is fixed and can be directly calculated to determine the detection position and size of the sealing line. The same grayscale value is used to inspect the product to be tested.

[0082] S4. Defect Analysis

[0083] The extracted defect and the surrounding area of ​​the defect are subjected to secondary binarization based on the grayscale value; the defect is re-extracted based on the result of the secondary binarization. Specifically, the extracted defect is analyzed for contour to obtain the defect's circumscribed rectangle. With the current rectangular frame as the center of a single unit, a rectangular frame of the same size is expanded in the upper, lower, left, and right directions, and the four corners are also filled in to form a nine-square grid. Within the area framed by the nine-square grid, a histogram grayscale value analysis is performed again to obtain the grayscale median, and the secondary binarization effect is performed again to obtain the accurately extracted defect, eliminating the influence of image unevenness.

[0084] In addition, the defect is subjected to pixel and / or size analysis to further extract valid defects. In this embodiment, the defect is subjected to pixel and size analysis, specifically: obtaining the circumscribed rectangle and area of ​​the defect outline; judging whether the pixels of the circumscribed rectangle and the defect outline meet the requirements of a valid defect, for example, if the length and width of the circumscribed rectangle are both greater than 2 pixels and the outline area is greater than 10 pixels, then it is extracted as a valid defect; if so, judging whether the size of the circumscribed rectangle and the area meets the control standard for defective products, for example, the control standard is defect length>0.015um, area>0.001mm 2 .

[0085] More preferably, a secondary analysis is performed on the defects within a certain value close to the card control standard, specifically: obtaining the maximum grayscale value, the minimum grayscale value and the grayscale mean of the defect; calculating the difference between the maximum grayscale value and the minimum grayscale value, recorded as the first difference, and the difference between the defect grayscale mean and the grayscale median of the block where the current defect is located, recorded as the second difference; if the first difference is greater than its corresponding first fixed value, and the second difference is also greater than its corresponding second fixed value, then an accurate secondary analysis is required. When a secondary analysis is required, use the defect grayscale mean + (defect grayscale mean - grayscale median of the block where the current defect is located) * Value, where Value represents the proportion, generally 0 or 10%, to re-extract the defects of the original image to recalculate the defect size; and then re-obtain the circumscribed rectangle and area of ​​the defect to obtain data closer to the actual size, thereby excluding the edge transition pixels of the defect.

[0086] S5. Data Recording

[0087] For each detected defect, record its location, the length and width of the defect's bounding rectangle, the defect's outline area, and other information. Ideally, defect information is recorded during each inspection. During product re-inspection, after a defect is detected, search the defect list for it. If a defect with the same location information is found, it will not be counted. This eliminates the influence of camera and lens contamination.

[0088] Based on the above method, dirt defects on the product surface can be detected accurately, quickly and stably; at the same time, the requirements for the overall grayscale consistency and uniformity of the image are reduced, and it can be compatible with the overall grayscale changes caused by material and processing differences in different products, and the detection accuracy of defects with smaller grayscale contrast differences and subtle defects is higher.

[0089] In addition, before executing step S1, it is also necessary to perform the step of collecting image data of the product to be tested. When collecting image data, the hardware light source is the lighting equipment in the whole system that cooperates with the camera to collect images. For each product, after the product is in place, turn on the light source first, then turn on the camera to collect images, and turn off the light source after the image collection is completed to obtain a processed image. Among them, different light source combinations can be used for different defects to collect different image effects. The hardware light source in this solution is a high-brightness, high-angle ring and square white light source that directly illuminates the product surface to highlight product defects. Since this solution has lower requirements for the grayscale consistency and uniformity of the overall image, it facilitates the installation and debugging of the hardware light source consistency and ensures the grayscale consistency of local areas in the product.

[0090] In conjunction with the CMOS chip surface defect detection method of the embodiment of the present invention, the CMOS chip surface defect detection system of the embodiment of the present invention is further introduced. The detection system includes an image acquisition module, a positioning module, a region division module, a block division module, a defect extraction module, a defect analysis module, and a data recording module. Among them, the defect extraction module includes a binarization processing unit and a defect extraction unit, and the defect analysis module includes an analysis unit and a processing unit. Specifically:

[0091] The image acquisition module is used to collect image data of the product to be tested. The collected image data of the product to be tested is used for subsequent operations such as area division, defect detection, and defect analysis.

[0092] The positioning module is used to calculate the positional deviation of the image data of the product to be tested. After the binarization unit performs binarization processing on the image data of the product to be tested based on the grayscale difference, the image area and non-image area of ​​the image data of the product to be tested are re-determined. The region division module performs edge detection and division of the detection area of ​​the image data of the product to be tested to determine the image area and non-image area.

[0093] The block division module divides the image area in the image data of the product to be tested and forms a plurality of detection blocks for defect detection.

[0094] The binarization unit performs binarization on each of the detection blocks and non-image areas in the image area based on grayscale values. The defect extraction unit is configured to extract defects from each of the detection blocks and non-image areas in the image area. Furthermore, for the initially extracted defects, the binarization unit performs a secondary binarization on the extracted defects and the areas surrounding them based on grayscale values. The defect extraction unit then re-extracts defects from the defects and their surrounding areas after the secondary binarization process.

[0095] The defect analysis module performs pixel and / or dimensional analysis on the extracted defects to further extract valid defects. Specifically, the analysis unit obtains the bounding rectangle and area of ​​the defect outline and determines whether the pixels and dimensions of the bounding rectangle, as well as the dimensions of the area, meet the valid defect and non-compliant product control standards. The processing unit performs a secondary analysis on defects within a certain value of the control standards to eliminate edge transition pixels of the defects and re-obtain the bounding rectangle and area of ​​the defects, thereby obtaining data closer to the actual size.

[0096] The data recording module records the location of the extracted defects, as well as the length and width of the defect's circumscribed rectangle, the defect contour area and other data.

[0097] Specifically, the implementation process of the functions and effects of each module and unit in this system is detailed in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0098] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any technician familiar with this profession can make some changes or modifications to equivalent embodiments of the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A CMOS chip surface defect detection method, characterized in that: The detection method comprises the following steps: Divide the image area in the image data of the product to be tested into multiple detection blocks; Performing binarization processing on each detection block according to the grayscale value; Extracting defects in each of the detection blocks according to the result of the binarization process; Performing pixel and / or size analysis on the defects to further extract valid defects; Selecting a fixed grayscale value for a non-image area of ​​the image data of the product to be tested; performing a binarization process on each of the non-image areas according to the comparison with the fixed grayscale value; Extracting defects in the non-image area according to a result of binarization processing; Perform pixel and / or size analysis on the defects to further extract valid defects, including: Obtain the circumscribed rectangle and area of ​​the defect outline; Determine whether the pixels of the circumscribed rectangle and the defect outline meet the requirements of a valid defect. If so, determine whether the sizes of the circumscribed rectangle and the area meet the control standard for defective products. Perform a secondary analysis on the defects within a certain value close to the control standard, specifically: obtain the maximum grayscale value, the minimum grayscale value, and the grayscale mean of the defect; calculate the difference between the maximum grayscale value and the minimum grayscale value, recorded as the first difference, and the difference between the defect grayscale mean and the grayscale median of the block where the current defect is located, recorded as the second difference; if the first difference is greater than its corresponding first fixed value, and the second difference is also greater than its corresponding second fixed value, then a precise secondary analysis is required. When a secondary analysis is required, use the defect grayscale mean + (defect grayscale mean - grayscale median of the block where the current defect is located) * Value, where Value represents the proportion, which is 0 or 10%, to re-extract the defects of the original image to recalculate the defect size.

2. The CMOS chip surface defect detection method according to claim 1, wherein: Before the step of dividing the image area, the detection method further includes the following steps: Performing binarization processing on the image data of the product to be tested according to the grayscale value; The image data of the product to be tested is subjected to edge detection processing on all four sides according to the binarization processing result, and a detection area is divided according to the edge detection result of all four sides. The detection area division result includes the image area and the non-image area.

3. The CMOS chip surface defect detection method according to claim 2, wherein: The detection method further comprises the steps of: Extracting the image area and obtaining the positioning center of the image data of the product to be tested; Calculating the position deviation of the image data of the product to be measured according to the positioning center; A positioning offset is performed according to the position deviation to redefine the image area and the non-image area.

4. The CMOS chip surface defect detection method according to claim 1, wherein: In the step of performing binarization processing on each of the detection blocks according to the grayscale value, specifically: Obtaining the grayscale median value of each detection block; Each detection block is binarized according to upper and lower set differences compared to the grayscale median value.

5. The CMOS chip surface defect detection method according to claim 1, wherein: The detection method further comprises the steps of: Performing secondary binarization processing on the extracted defect and the area surrounding the defect according to the grayscale value; Defects are re-extracted according to the result of the secondary binarization process.

6. The CMOS chip surface defect detection method according to claim 1, wherein: Before performing binarization processing on the non-image area, the method further includes the following steps: Eliminate the influence of edge lines in the non-image area of ​​the product to be tested.

7. A CMOS chip surface defect detection system, characterized in that: Used to perform the method according to any one of claims 1 to 6, comprising: Image acquisition module, used to collect image data of the product to be tested; A block division module is used to divide the image area in the image data of the product to be tested and form a plurality of detection blocks for defect detection; The defect extraction module includes: a binarization processing unit for performing binarization processing on each of the detection blocks according to the grayscale value; a defect extraction unit for extracting defects in each detection block according to the result of the binarization processing; A defect analysis module, configured to perform pixel and / or size analysis on the extracted defects to further extract valid defects; The defect analysis module includes: An analysis unit, configured to obtain a circumscribed rectangle and an area of ​​the defect outline, and determine whether the pixels and / or sizes of the circumscribed rectangle and the area meet the valid defect and defective product control standards; The processing unit is configured to analyze and process the defects within a certain value close to the card control standard, and redetermine the bounding rectangle and area of ​​the defect after excluding the influence of the edge transition pixels of the defect, wherein the analysis and processing specifically comprises: obtaining the maximum grayscale value, the minimum grayscale value, and the grayscale mean of the defect; calculating the difference between the maximum grayscale value and the minimum grayscale value, recorded as a first difference, and the difference between the defect grayscale mean and the grayscale median of the block where the current defect is located, recorded as a second difference; if the first difference is greater than its corresponding first fixed value, and the second difference is also greater than its corresponding second fixed value, then a precise secondary analysis is required; when secondary analysis is required, using the defect grayscale mean + (defect grayscale mean - grayscale median of the block where the current defect is located) * Value, where Value represents a percentage, which is 0 or 10%, to re-extract the defect from the original image to recalculate the defect size; and then re-obtaining the bounding rectangle and area of ​​the defect to obtain data closer to the actual size, thereby excluding the edge transition pixels of the defect.

8. The CMOS chip surface defect detection system according to claim 7, wherein: The binarization processing unit is further configured to perform binarization processing on the image data of the product to be tested according to the grayscale value; The detection system also includes a region division module, which performs edge detection processing on the image data of the product to be tested based on the binarization processing result, and divides the detection area according to the edge detection result of the four edges. The detection area division result includes the image area and the non-image area.

9. The CMOS chip surface defect detection system according to claim 7, wherein: The detection system further includes a positioning module, which is used to calculate the position deviation of the image data of the product to be tested and re-determine the image area and non-image area of ​​the image data of the product to be tested.

10. The CMOS chip surface defect detection system according to claim 7, wherein: The binarization processing unit is further used to perform binarization processing on the non-image area of ​​the image data of the product to be tested according to the grayscale value, and the defect extraction unit is further used to extract defects in the non-image area according to the processing result of the binarization processing unit.

11. The CMOS chip surface defect detection system according to claim 7, wherein: The binarization processing unit is further used to perform secondary binarization processing on the extracted defect and the defect surrounding area according to the grayscale value, and the defect extraction unit is further used to re-extract defects from the defect and the surrounding area after the secondary binarization processing.

Images