CN118298203A - Defect clustering method, system and equipment - Google Patents

Abstract

The defect clustering method, the defect clustering system and the defect clustering equipment provided by the invention are used for identifying defects of detected images and determining original defect blocks and first position information of the original defect blocks; mapping the first position information of each original defect block in the detected image according to a preset condition to obtain second position information of each original defect block after mapping; and determining target defect blocks to be clustered according to the second position information of each original defect block, and further clustering the target defect blocks. According to the invention, the first position information of each original defect block is mapped, so that the second position information of each original defect block is more compact, the target defect block needing to be clustered can be more accurately determined, the data volume of clustering is reduced, and the defect detection efficiency is improved.

Description

Defect clustering method, system and equipment

Technical Field

The invention relates to the field of defect detection, in particular to a defect clustering method, system and equipment.

Background

Defect detection generally refers to detection of surface defects of an article, wherein the surface defects are detected by adopting advanced machine vision detection technology, such as spots, pits, scratches, color differences, defects and the like on the surface of a workpiece.

For products such as semiconductors, defects (such as scratches) are usually very tiny, and in a product photo shot by a camera, the defects are not obvious in the photo, and the same defect is easily identified into a plurality of defects, so that the defects are generally classified in a clustering mode in the prior art, iteration is needed in a conventional clustering mode, when the number of the defects is large, the data processing amount of clustering processing is greatly increased, time consumption is greatly increased, and the defect detection efficiency is reduced.

Disclosure of Invention

The application provides a defect clustering method, a defect clustering system and defect clustering equipment, and aims to improve defect detection efficiency.

A defect clustering method, comprising:

Acquiring a detection image of a piece to be detected;

Performing defect identification on the detected image, and determining an original defect block and first position information of the original defect block, wherein each original defect block comprises one or more pixels;

Mapping the first position information of each original defect block in the detection image according to a preset condition to obtain second position information of each original defect block after mapping;

determining target defect blocks to be clustered according to the second position information of each original defect block;

And clustering the target defect blocks.

In the defect clustering method provided in an embodiment, the first location information includes a first location coordinate, and the second location information includes a second location coordinate; the mapping processing is performed on the first position information of each original defect block in the detected image according to a preset condition to obtain second position information after mapping each original defect block, including:

mapping the first position coordinates of each original defect block in the detection image according to a preset value to obtain corresponding second position coordinates; the preset value is smaller than 1.

In the defect clustering method provided in an embodiment, the first location information includes a first location coordinate, and the second location information includes a second location coordinate; the mapping processing is performed on the first position information of each original defect block in the detected image according to a preset condition to obtain second position information after mapping each original defect block, including:

reducing the size of the detection image according to a preset value to obtain a mapping image;

mapping each original defect block in the detection image to the mapping image according to the preset value, and obtaining a second position coordinate of each original defect block in the mapping image; the preset value is smaller than 1.

In the defect clustering method provided in an embodiment, the determining, according to the second location information of each original defect block, a target defect block to be clustered includes:

And acquiring original defect blocks with the same second position coordinates according to the second position coordinates of the original defect blocks, determining the original defect blocks as the target defect blocks, and acquiring original defect blocks with the distances between the second position coordinates of the original defect blocks being more than 0 and less than a first preset distance, and determining the original defect blocks as intermediate defect blocks.

In the defect clustering method provided in an embodiment, the first preset distance isA pixel distance; or the first preset distance is 2 pixel distances.

In the defect clustering method provided in an embodiment, the determining, according to the second location information of each original defect block, the target defect block to be clustered further includes:

And determining whether the distance between the intermediate defect blocks is smaller than or equal to a second preset distance according to the first position coordinates of the intermediate defect blocks, and if the distance between the intermediate defect blocks is smaller than or equal to the second preset distance, determining the intermediate defect blocks as the target defect blocks.

In the defect clustering method provided in an embodiment, the performing defect recognition on the detected image to determine an original defect block and first location information of the original defect block includes:

And carrying out defect identification on the detected image, determining the original defect block and corner pixels or center point pixels of the original defect block, and taking the position coordinates of the corner pixels or the center point pixels of the original defect block as first position coordinates of the original defect block.

An embodiment provides a defect clustering system, comprising:

The image acquisition module is used for acquiring a detection image of the to-be-detected piece;

An image processing module for:

Performing defect identification on the detected image, and determining an original defect block and first position information of the original defect block, wherein each original defect block comprises one or more pixels;

Mapping the first position information of each original defect block in the detection image according to a preset condition to obtain second position information of each original defect block after mapping;

determining target defect blocks to be clustered according to the second position information of each original defect block;

And clustering the target defect blocks.

An embodiment provides a defect detection apparatus including:

a memory for storing a program;

A processor for implementing the method of any of claims 1-7 by executing a program stored in the memory.

An embodiment provides a computer-readable storage medium having stored thereon a program executable by a processor to implement a method as described above.

According to the defect clustering method, system and equipment of the embodiment, defect identification is carried out on the detected image, and an original defect block and first position information of the original defect block are determined; mapping the first position information of each original defect block in the detected image according to a preset condition to obtain second position information of each original defect block after mapping; and determining target defect blocks to be clustered according to the second position information of each original defect block, and further clustering the target defect blocks. According to the invention, the first position information of each original defect block is mapped, so that the second position information of each original defect block is more compact, the target defect block needing to be clustered can be more accurately determined, the data volume of clustering is reduced, and the defect detection efficiency is improved.

Drawings

FIG. 1 is a flowchart of an embodiment of a defect clustering method according to the present invention;

FIG. 2 is a simplified schematic diagram of a detected image in the defect clustering method provided by the invention;

FIG. 3 is a flowchart for determining that the second position coordinates are adjacent in the defect clustering method provided by the present invention;

FIG. 4 is a map image corresponding to the detected image of FIG. 2;

FIG. 5 shows a schematic diagram of 8-neighborhood and 4-neighborhood positional relationships;

fig. 6 is a block diagram of a defect clustering system according to an embodiment of the present invention.

Detailed Description

The application will be described in further detail below with reference to the drawings by means of specific embodiments. Wherein like elements in different embodiments are numbered alike in association. In the following embodiments, numerous specific details are set forth in order to provide a better understanding of the present application. However, one skilled in the art will readily recognize that some of the features may be omitted, or replaced by other elements, materials, or methods in different situations. In some instances, related operations of the present application have not been shown or described in the specification in order to avoid obscuring the core portions of the present application, and may be unnecessary to persons skilled in the art from a detailed description of the related operations, which may be presented in the description and general knowledge of one skilled in the art.

Furthermore, the described features, operations, or characteristics of the description may be combined in any suitable manner in various embodiments. Also, the various steps or acts in the method descriptions may be interchanged or adjusted in a manner apparent to those skilled in the art. Thus, the various orders in the description and drawings are for clarity of description of only certain embodiments, and are not meant to be required orders unless otherwise indicated.

The numbering of the components itself, e.g. "first", "second", etc., is used herein merely to distinguish between the described objects and does not have any sequential or technical meaning. The term "coupled" as used herein includes both direct and indirect coupling (coupling), unless otherwise indicated.

In the defect detection field, it is necessary to cluster each defect block in the detected image, so as to determine which defect blocks belong to the same defect and which do not belong to the same defect, thereby realizing defect detection. However, the defect detection in the semiconductor field is microscopic detection, the precision requirement is high, each defect block is pixel-level, the number is very large, and the defect detection consumes a long time. According to the invention, the positions of the defect blocks are mapped and reduced in advance, and the reduced position relationship is more compact, so that a part of defect blocks can be better screened out for clustering, all the defect blocks are not required to be clustered, the operation amount of clustering is reduced, the defect detection efficiency is improved, and the method is described in detail by some embodiments.

As shown in fig. 1, the defect clustering method provided by the invention comprises the following steps:

And step 1, acquiring a detection image of the to-be-detected piece. The detected image contains defect information, for example, it is known whether each pixel on the detected image is a defective pixel or a normal pixel. The detection image may be a binarized image. Taking a piece to be detected as a chip as an example, a camera is controlled to shoot a silicon substrate to obtain an original image, the original image generally comprises a plurality of chip images, each chip image is extracted from the original image, and the chip images can be independently shot to obtain the chip images. The types of chips are known, and standard images and deviation thresholds are preset for the chips of each type. The standard image may be a manually confirmed defect-free chip image. The same position of the chip image and the standard image can be provided with a positioning mark, and the positioning mark is used for positioning pixels on the image. The chip image may be subjected to differential processing by using the standard image to obtain a differential image, for example, after the chip image and the standard image are positioned by using the positioning mark, a difference value (absolute value) of pixel values of the same pixel on the chip image and the standard image is calculated, the difference value is used as a pixel value of a corresponding pixel on the differential image, and the differential image is subjected to binarization processing by using a preset deviation threshold as a binarization threshold to obtain the detection image. In other words, whether the pixel value (difference value) of each pixel on the differential image exceeds a preset deviation threshold value is respectively judged, and if the pixel value of the pixel exceeds the deviation threshold value, the pixel is determined to be a defective pixel; if the pixel value of the pixel does not exceed the deviation threshold value, the pixel value is considered to belong to a normal pixel, and the pixel value is set to 0 or 255, so that the pixel is better distinguished from the defective pixel. After each pixel is judged and processed in this process, the differential image becomes a detection image. The detection image obtained in this way only contains defective pixels, so that the operation amount of the subsequent steps is greatly reduced. Of course, in other embodiments, the detected image may be obtained in other specific manners, as long as it is known whether each pixel on the detected image is a defective pixel.

Step 2, performing defect identification on the detected image, and determining an original defect block and first position information of the original defect block, wherein each original defect block comprises one or more pixels. For example, the connected domain is found for each defective pixel on the detected image, and each connected domain obtained is an original defective block. In the detected image shown in fig. 2, each cell represents one pixel, the pixel with gray scale is a defective pixel, and the white pixel is a normal pixel. Solving connected domain of each defective pixel on the detected image to obtain four original defective blocks: 1. 2, 3 and 4.

The individual connected domains may also be filtered, such as by area, aspect ratio, rotating rectangle, or rotating angle, etc.

The first location information of the original defective block includes first location coordinates. The first position coordinates of each original defect block in the detected image may be determined according to the position of each original defect block in the detected image. Specifically, the pixel of the upper left corner of the detected image may be taken as a starting point (coordinate origin), each pixel uses the starting point as a position reference to obtain the position coordinate of each pixel in the detected image, and a single pixel may be taken as a coordinate unit, so that the position coordinate of the pixel includes the number of pixels differing from the starting point (0, 0) in the transverse direction and the number of pixels differing from the starting point in the longitudinal direction, the first position coordinate of the original defect block 2 in fig. 2 is (4, 4), and since the position relationship of each pixel in the image is known, the operation amount can be simplified when determining the position coordinate of each original defect block and the subsequent neighborhood judgment. And further, determining the corner (such as the upper left corner) pixel or the center point pixel of each original defect block, and taking the position coordinates of the corner pixel or the center point pixel of the original defect block as the first position coordinates of the original defect block. The present embodiment will be described taking the position coordinates of the center point pixel as the first position coordinates of the original defective block as an example.

For convenience of distinction, each original defective block may be numbered or the like (1, 2,3, and 4 in fig. 2), or an circumscribed rectangle of the original defective block may be obtained according to the position of the pixel included in the original defective block, or the like.

And step 3, mapping the first position information of each original defect block in the detected image according to preset conditions, such as mapping reduction processing, to obtain second position information of each original defect block after mapping. The preset condition may be a reduction magnification, for example, the first location information of each original defect block in the detected image is mapped according to a preset value, for example, the mapping reduction process is performed, so as to obtain second location information after mapping each original defect block. The preset value is greater than 0 and less than 1, and can be specifically set according to the needs of users.

In this embodiment, the second position information may also include second position coordinates. That is, the first position coordinates of each original defect block in the detected image are reduced according to a preset value, and corresponding second position coordinates are obtained. The preset value may be the reciprocal of a preset cluster radius value, specifically, the reciprocal of the preset cluster radius value is taken as the multiplying power to reduce and round the first position coordinates of each original defect block to obtain corresponding second position coordinates, or the reciprocal of the preset cluster radius value (preset value) is multiplied with the first position coordinates of each original defect block and rounded to obtain corresponding second position coordinates. The rounding may be upward rounding or downward rounding, and this embodiment is described by taking upward rounding as an example. As shown in fig. 2, the size of the detected image is 12×8, i.e., the position coordinates of the lower right corner point pixel are (12, 8). Taking the clustering radius value rad=2 as an example, the center coordinate (first position coordinate) of the original defect block 1 is (11, 1), the mapped coordinate is (5.5,0.5), the final coordinate (second position coordinate) after rounding up is (6, 1), and the like, the second position coordinates after mapping all the original defect blocks can be calculated.

In other embodiments, the second location information mapped by each original defect block may be obtained as follows:

And reducing the size of the detection image according to a preset value to obtain a mapping image. As shown in fig. 4. The size of the detected image is 6*4 after the size is reduced by 1/2 of the reciprocal of the cluster radius value. Mapping each original defect block in the detection image to a mapping image according to a preset value, and obtaining a second position coordinate of each original defect block in the mapping image. Specifically, mapping each pixel point in the detection image onto the mapping image can be realized by adopting the way that the position coordinates are reduced and rounded according to the reciprocal of the clustering radius value, so that the position coordinates in the mapping image after each pixel point in the detection image is mapped onto the mapping image are obtained. The mapping image comprises mapping defect blocks corresponding to the original defect blocks, namely after each pixel of the original defect blocks is mapped onto the mapping image according to a preset value, the pixels on the mapping image form the corresponding mapping defect blocks, and therefore second position coordinates of each original defect block in the mapping image are obtained.

And 4, determining target defect blocks to be clustered according to the second position information of each original defect block. Specifically, an original defect block having the same second position coordinates may be acquired based on the second position coordinates of each original defect block and determined as a target defect block, and an original defect block adjacent to the second position coordinates may be acquired and determined as an intermediate defect block.

Wherein, as shown in fig. 3, the original defect blocks adjacent to the second position coordinates are obtained according to the second position coordinates of each original defect block and are determined as intermediate defect blocks, which can be realized by the following steps:

And step 41, judging whether other mapping defect blocks exist in the neighborhood of each mapping defect block in the mapping image according to the position coordinates of the mapping defect block on the mapping image. For example, it is sequentially judged whether or not other mapping defect blocks exist in the neighborhood of the mapping defect blocks 1 to 4 in fig. 4.

And step 42, if other mapping defect blocks exist in the neighborhood of the mapping defect block, determining the original defect block corresponding to each mapping defect block in the mapping defect block and the neighborhood thereof as an intermediate defect block. The neighborhood may be an 8 neighborhood or a 4 neighborhood, as shown in fig. 5, and for pixel 13, the 8 neighborhood is pixel 7, 8, 9, 12, 14, 17, 18, and 19. The 4 neighbors of pixel 13 are pixels 8, 12, 14 and 18.

In steps 41 and 42, specifically, the distance between the second position coordinates in each original defect block may be obtained, and the original defect block whose distance between the second position coordinates is greater than 0 and smaller than the first preset distance is determined as the intermediate defect block. The first preset distance can be set according to the requirement, for example, the first preset distance isThe pixel distance corresponds to a neighborhood of 4. Taking the pixel point 13 as an example, the distance between the pixel point 13 and the pixel point 13 is more than 0 and less thanThe pixels at a pixel distance are 8, 12, 14 and 18. The first preset distance may also be a 2 pixel distance, corresponding to 8 neighbors. Taking the pixel 13 as an example, the pixels having a distance from the pixel 13 greater than 0 and less than 2 pixels have 7,8, 9, 12, 14, 17, 18 and 19. This embodiment will be described with 8 neighbors as an example. For example, if the position coordinates of the pixels at the center point of the original defect block are used as the first position coordinates, there are no adjacent mapping defect blocks in the neighborhood of the corresponding 4 mapping defect blocks in fig. 4, and the second position coordinates of the original defect blocks 1-4 in fig. 2 are not adjacent to each other. If the position coordinates of the upper left corner pixel of the original defective block are taken as the first position coordinates, there is no adjacent mapped defective block in the neighborhood of the mapped defective blocks 1 and 4 in fig. 4, i.e., the second position coordinates of the original defective blocks 1 and 4 in fig. 2 are not adjacent to each other. Adjacent mapping defect blocks 3 are located in the neighborhood of the mapping defect blocks 2, and conversely, adjacent mapping defect blocks 2 are located in the neighborhood of the mapping defect blocks 3, so that the original defect blocks 2 and 3 are adjacent to each other in the second position coordinates, and the original defect blocks 2 and 3 are determined as intermediate defect blocks. The neighborhood judgment is carried out on the mapping image, so that the method has small operand and high efficiency, and is very suitable for defect detection of semiconductor products.

By the method, whether the second position coordinate of each original defect block is adjacent to or the same as the second position coordinates of other original defect blocks can be obtained. So that the original defective blocks having the same second position coordinates can all be determined as target defective blocks. In order to further reduce the calculation amount, the original defect blocks having the same second position coordinates may be classified into a class, that is, the original defect blocks having the same second position coordinates are considered to belong to the same defect, specifically, the original defect blocks having the same second position coordinates may be combined into one and the combined original defect block may be determined as a target defect block, and the first position coordinates of the target defect block may be the position coordinates of a center point pixel or a corner point pixel of an circumscribed rectangle formed by the original defect blocks having the same second position coordinates. The method classifies part of original defect blocks in the step, and reduces the data volume of the next clustering step.

For intermediate defect blocks (original defect blocks adjacent to the second position coordinates), it is not possible to directly determine whether or not the original defect blocks belong to the same defect, so that the distance between the intermediate defect blocks (the distance between the intermediate defect blocks on the detected image) can be determined according to the first position coordinates of the intermediate defect blocks, whether the distance is less than or equal to a second preset distance is determined, and if the distance is less than or equal to the second preset distance, the intermediate defect block is determined as the target defect block. The second preset distance may be set as required, and in this embodiment, the second preset distance is a cluster radius value. Intermediate defect blocks are screened through the clustering radius value, so that the number of defect blocks needing to be clustered subsequently is reduced.

For the original defect blocks with different and non-adjacent second position coordinates, the original defect blocks are far apart from each other on the detected image, and the original defect blocks can be not clustered, in other words, for the original defect blocks with different and non-adjacent second position coordinates, the original defect blocks can be considered to belong to different defects. Similarly, if the distance between the intermediate defect blocks on the detected image exceeds the second preset distance, the intermediate defect blocks are far apart from each other, and the intermediate defect blocks can not be clustered.

And step 5, clustering the target defect blocks. The target defect blocks may be clustered by various known clustering algorithms, and each cluster (class) obtained by clustering is a defect. Therefore, the clustering is performed according to the method, and the defect blocks are screened, so that the data volume of the clustering is reduced, and the defect detection efficiency is improved. In this embodiment, distance clustering is taken as an example to describe, and specifically, clustering is performed on target defect blocks based on a preset cluster radius value. Thus, the steps shown in fig. 1 are performed around the cluster radius value set by the user, and the final result also meets the accuracy corresponding to the cluster radius value.

Clustering the target defect blocks based on the preset cluster radius values may be performed as follows:

Respectively taking each target defect block as a clustering center, and carrying out clustering operation on target defect blocks with second position coordinates adjacent to the second position coordinates of the clustering center (the distance between the second position coordinates of the clustering center is smaller than the first preset distance) according to a preset clustering radius value to obtain a clustering result; each class in the clustering result represents a defect. Of course, other modes may be adopted, and the target defect blocks are mostly formed by original defect blocks adjacent to the second position coordinates, so that the mapping defect blocks corresponding to the target defect blocks on the mapping image are mostly adjacent, the connected domain of the mapping defect blocks may be obtained, and clustering operation may be performed on the original defect blocks (target defect blocks) corresponding to the mapping defect blocks in each connected domain according to a preset clustering radius value, so as to obtain a clustering result. Both of these are essentially identical.

After the defects on the detected images are classified in the above manner, the method can also be applied, for example, the defects are marked with positions and/or shapes on the detected images or the associated images, and the positions and the shapes of the defects on the to-be-detected piece are displayed on a display interface of a display, so that a user is prompted. Thus, on-line defect detection can be performed, and the method is very efficient. The image associated with the detection image may be an original image for generating the detection image, such as a chip image, or an image generated from the detection image.

Of course, the clustering result is obtained by the method, that is, after the original defect blocks contained in each defect are obtained, the number of the defects can be counted and displayed on a display interface of a display, so that a user is prompted of the defect condition of the current to-be-detected piece, and further fine detection can be performed according to quality inspection standards.

Based on the defect clustering method, the invention also provides a defect clustering system, as shown in fig. 6, comprising: an image acquisition module 10 and an image processing module 20.

The image acquisition module 10 is configured to acquire a detection image of a piece to be detected, and may be acquired from an external device, or may control a camera to capture a silicon substrate, so as to obtain an original image, and further process the original image to obtain a detection image. That is, step 1 of the above embodiment may be completed by the image acquisition module 10, and specific processes are shown in step 1, which is not described herein.

The image processing module 20 is configured to perform defect recognition on the detected image, and determine an original defect block and first location information of the original defect block. For example, the image processing module 20 obtains connected domains for each defective pixel on the detected image, and each obtained connected domain is an original defective block; the individual connected domains may also be filtered, such as by area, aspect ratio, rotating rectangle, or rotating angle, etc.

The first location information of the original defective block includes first location coordinates. The image processing module 20 may determine the first position coordinates of each original defect block in the detected image according to the position of each original defect block in the detected image. That is, step 2 of the above embodiment may be completed by the image processing module 20, and specific processes are shown in step 2, which is not described herein.

The image processing module 20 further performs mapping processing on the first position information of each original defect block in the detected image according to a preset condition, so as to obtain second position information mapped by each original defect block. The preset condition may be a reduction magnification, for example, the image processing module 20 performs mapping processing on the first location information of each original defect block in the detected image according to a preset value, for example, performing mapping reduction processing, to obtain second location information after mapping each original defect block. That is, the step 3 in the above embodiment may be completed by the image processing module 20, and the specific process is referred to as the step 3, which is not described herein.

The image processing module 20 also determines target defect blocks to be clustered according to the second location information of each original defect block. Specifically, the image processing module 20 may acquire the original defect blocks having the same second position coordinates according to the second position coordinates of each of the original defect blocks and determine the same as the target defect block, and acquire the original defect blocks having a distance between the second position coordinates of each of the original defect blocks greater than 0 and less than the first preset distance and determine the same as the intermediate defect block. That is, step 4 of the above embodiment may be completed by the image processing module 20, and the specific process is referred to as step 4, which is not described herein.

The image processing module 20 further clusters the target defect blocks, and may use various known clustering algorithms to cluster each target defect block, where each cluster (class) obtained by the clustering is a defect. For example, the target defective blocks are clustered based on a preset cluster radius value. Specifically, the image processing module 20 may perform clustering operation on target defect blocks adjacent to the second position coordinate of the clustering center according to a preset clustering radius value by using each target defect block as a clustering center, so as to obtain a clustering result; each class in the clustering result represents a defect. Of course, other manners may be adopted, for example, the image processing module 20 may calculate the connected domain for the mapping defect block, and perform clustering operation on the original defect block (target defect block) corresponding to the mapping defect block in each connected domain according to a preset cluster radius value, so as to obtain a clustering result. Both of these are essentially identical.

After the defects on the detected images are classified in the above manner, the method can be applied, for example, the defect clustering system further comprises a display module. The image processing module 20 marks the position and/or shape of the defect on the detected image or its associated image, and the detected image marked with the position and/or shape of the defect or its associated image is displayed by the display module. The image processing module 20 may also count the number of defects, and the number of defects is displayed by the display module.

The invention also provides defect detection equipment which comprises a memory and a processor. Wherein the memory is used for storing programs. The processor is configured to implement the method shown in fig. 1 by executing a program stored in the memory. For example, the processor includes the image acquisition module 10 described above and the image processing module 20 described above. Since the method of executing the method shown in fig. 1 by the processor is described in detail in the foregoing embodiments, details are not described herein.

Those skilled in the art will appreciate that all or part of the steps of the various methods in the above embodiments may be implemented by a program to instruct related hardware, and the program may be stored in a computer readable storage medium, where the storage medium may include: read-only memory, random access memory, magnetic or optical disk, etc.

The foregoing description of the invention has been presented for purposes of illustration and description, and is not intended to be limiting. Several simple deductions, modifications or substitutions may also be made by a person skilled in the art to which the invention pertains, based on the idea of the invention.

Claims (10)

1. A defect clustering method, comprising:

Acquiring a detection image of a piece to be detected;

Performing defect identification on the detected image, and determining an original defect block and first position information of the original defect block, wherein each original defect block comprises one or more pixels;

Mapping the first position information of each original defect block in the detection image according to a preset condition to obtain second position information of each original defect block after mapping;

determining target defect blocks to be clustered according to the second position information of each original defect block;

And clustering the target defect blocks.

2. The defect clustering method of claim 1, wherein the first location information comprises first location coordinates and the second location information comprises second location coordinates; the mapping processing is performed on the first position information of each original defect block in the detected image according to a preset condition to obtain second position information after mapping each original defect block, including:

mapping the first position coordinates of each original defect block in the detection image according to a preset value to obtain corresponding second position coordinates; the preset value is smaller than 1.

3. The defect clustering method of claim 1, wherein the first location information comprises first location coordinates and the second location information comprises second location coordinates; the mapping processing is performed on the first position information of each original defect block in the detected image according to a preset condition to obtain second position information after mapping each original defect block, including:

reducing the size of the detection image according to a preset value to obtain a mapping image;

mapping each original defect block in the detection image to the mapping image according to the preset value, and obtaining a second position coordinate of each original defect block in the mapping image; the preset value is smaller than 1.

4. A defect clustering method as claimed in claim 2 or 3, wherein said determining the target defect blocks to be clustered based on the second location information of each original defect block comprises:

And acquiring original defect blocks with the same second position coordinates according to the second position coordinates of the original defect blocks, determining the original defect blocks as the target defect blocks, and acquiring original defect blocks with the distances between the second position coordinates of the original defect blocks being more than 0 and less than a first preset distance, and determining the original defect blocks as intermediate defect blocks.

5. The defect clustering method of claim 4, wherein the first predetermined distance isA pixel distance; or the first preset distance is 2 pixel distances.

6. The defect clustering method of claim 5, wherein determining the target defect blocks to be clustered based on the second location information of each original defect block further comprises:

And determining whether the distance between the intermediate defect blocks is smaller than or equal to a second preset distance according to the first position coordinates of the intermediate defect blocks, and if the distance between the intermediate defect blocks is smaller than or equal to the second preset distance, determining the intermediate defect blocks as the target defect blocks.

7. The defect clustering method of claim 1, wherein performing defect recognition on the detected image to determine an original defect block and first location information of the original defect block comprises:

And carrying out defect identification on the detected image, determining the original defect block and corner pixels or center point pixels of the original defect block, and taking the position coordinates of the corner pixels or the center point pixels of the original defect block as first position coordinates of the original defect block.

8. A defect clustering system, comprising:

The image acquisition module is used for acquiring a detection image of the to-be-detected piece;

An image processing module for:

Performing defect identification on the detected image, and determining an original defect block and first position information of the original defect block, wherein each original defect block comprises one or more pixels;

Mapping the first position information of each original defect block in the detection image according to a preset condition to obtain second position information of each original defect block after mapping;

determining target defect blocks to be clustered according to the second position information of each original defect block;

And clustering the target defect blocks.

9. A defect inspection apparatus, comprising:

a memory for storing a program;

A processor for implementing the method of any of claims 1-7 by executing a program stored in the memory.

10. A computer readable storage medium, characterized in that the medium has stored thereon a program executable by a processor to implement the method of any one of claims 1 to 7.