Chip support appearance defect detection template matching and positioning method and device

Abstract

A chip support appearance defect detection template matching and positioning method and device, the method comprising: 1) defining an image of a chip support template, positioning the product edge and performing 2D Euclidean transformation on the image to obtain a corrected image; 2) defining the corrected image of the template, constructing a template group and extracting a 1D fingerprint of the template group; 3) detecting a target image of the chip support, positioning the product edge and performing 2D Euclidean transformation on the image to obtain a target corrected alignment image; 4) extracting a target 1D fingerprint from the target corrected alignment image; 5) matching the 1D fingerprint of the template group with the target 1D fingerprint to obtain a template group offset; and 6) determining the matched template information in the target corrected alignment image according to the template group offset and the template information in the template group. The present application realizes accurate and efficient chip support appearance detection template matching and positioning.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of machine vision defect detection, and particularly relates to a chip support appearance defect detection template matching and positioning method and device. BACKGROUND

[0002] The chip support, also known as a lead frame, is a chip carrier of a semiconductor, is a key structural part for forming an electrical circuit by realizing electrical connection between an external circuit (PCB) and an internal circuit of a chip through bonding wires. The lead frame plays a role of a bridge for connecting with external wires, and is used in most semiconductors, is an important basic material in the electronic information industry, and is mainly produced by a die stamping method and an etching method.

[0003] Chip support appearance defect detection usually adopts a method based on modeling and comparison, in which template matching and positioning is a key point, and its implementation greatly determines the accuracy and efficiency of chip support appearance defect detection. Due to the shape characteristics of chip support products and factors of the production process environment, the current chip support appearance defect detection template matching and positioning method faces great challenges in accuracy and efficiency:

[0004] 1. Product shaking and tilting lead to template matching and positioning errors: on the one hand, the punched products are accompanied by significant shaking while constantly moving, and tilting and lateral movement easily occur during detection shooting, so that a simple template matching method will fail or make errors. On the other hand, different product units of some models of chip supports are very similar, only with slight differences, and template matching is prone to confusion and errors, and the problem is more serious when the products are tilted.

[0005] 2. Large search range of product template matching and positioning leads to low efficiency: in the chip support appearance defect detection scene, template matching and positioning often need to search a large 2D range, and in an extreme case, the product angle search dimension is considered to become 3D search due to product tilting. Under the conditions of relatively fast production line speed and relatively high detection accuracy requirement, the current product template matching and positioning method has the problem of insufficient efficiency.

[0006] The above problems seriously hinder the improvement of accuracy and efficiency of chip support appearance defect detection, and bring an unbearable cost burden to the production and detection of chip supports. Therefore, it is necessary to propose a new chip support appearance defect detection template matching and positioning method to solve the above two outstanding problems. SUMMARY

[0007] The present application provides a chip support appearance defect detection template matching and positioning method and device to solve the problems of insufficient accuracy and efficiency of current chip support appearance defect detection template matching and positioning, and realizes accurate and efficient template matching and positioning.

[0008] The first aspect of the present application relates to a chip carrier appearance defect detection template matching and positioning method for detecting appearance defects of a chip carrier, specifically comprising the following steps:

[0009] Step S1: defining an image of a chip carrier template, positioning a product edge, and performing 2D Euclidean transformation on the image to obtain a corrected image;

[0010] Step S2: defining a corrected image of the template, constructing a template group, and extracting a 1D fingerprint of the template group;

[0011] Step S3: detecting a target image of the chip carrier, positioning a product edge, and performing 2D Euclidean transformation on the image to obtain a target corrected alignment image;

[0012] Step S4: extracting a target 1D fingerprint from the target corrected alignment image;

[0013] Step S5: matching the 1D fingerprint of the template group with the target 1D fingerprint to obtain a template group offset;

[0014] Step S6: determining the matching template information in the target corrected alignment image according to the template group offset and the template information in the template group.

[0015] According to the above technical solution, the specific steps of positioning the product edge based on the chip carrier template definition image in step S1 and performing 2D Euclidean transformation on the image to obtain a corrected image are as follows:

[0016] Step S1.1: cutting a detection area image from the template definition image, respectively analyzing the pixel values of the upper, lower, left, and right four boundaries of the detection area image, and determining the standard product positioning edge type and quantity;

[0017] Step S1.2: for each standard product positioning edge, starting from the boundary pixel of the detection area image, searching for product edge pixels from the outside to the inside to obtain a standard product positioning edge pixel set;

[0018] Step S1.3: for each standard product positioning edge pixel set, fitting a straight line to obtain a standard product positioning edge straight line;

[0019] Step S1.4: determining a correction transformation according to the standard product positioning edge straight line, applying the correction transformation to the template definition image to obtain a template definition corrected image.

[0020] According to the above technical solution, the specific steps of cutting a detection area image from the template definition image in step S1.1, respectively analyzing the pixel values of the upper, lower, left, and right four boundaries of the detection area image, and determining the standard product positioning edge type and quantity are as follows:

[0021] Step S1.1.1: cutting a detection area image from the template definition image;

[0022] Step S1.1.2: respectively performing statistical analysis on pixel values of upper, lower, left and right four boundaries of the detection area image to obtain mean values and standard deviations of the pixel values of the upper, lower, left and right four boundaries;

[0023] Step S1.1.3: determining whether the product positioning edge is single or double and which type of upper edge, lower edge, left edge or right edge each product positioning edge is according to the mean values and standard deviations of the pixel values of the upper, lower, left and right four boundaries of the detection area image.

[0024] According to the above technical solution, the specific steps of determining the correction transformation according to the standard product positioning edge straight line in step S1.4 are as follows:

[0025] Step S1.4.1: determining a correction transformation rotation center point according to the standard product positioning edge straight line;

[0026] Step S1.4.2: determining a correction transformation rotation angle according to the direction of the standard product positioning edge straight line;

[0027] Step S1.4.3: rotating the template definition image according to the rotation center point and the rotation angle of the correction transformation to obtain a template definition correction image.

[0028] According to the above technical solution, the specific steps of constructing a template group and extracting a template group 1D fingerprint from the template definition correction image in step S2 are as follows:

[0029] Step S2.1: performing 1D self-matching on the template definition correction image to determine a product unit image size and a repeating mode;

[0030] Step S2.2: extracting a template image from the template definition correction image according to the product unit image size and the repeating mode, and splicing the template image to form a template group according to the product unit repeating mode;

[0031] Step S2.3: reducing the spliced template group image to obtain a template group 1D fingerprint.

[0032] According to the above technical solution, the specific steps of locating a product edge and performing 2D Euclidean transformation on the chip support detection target image to obtain a target correction alignment image in step S3 are as follows:

[0033] Step S3.1: searching product edge pixel points from the detection area boundary pixels outwardly and inwardly according to the product positioning edge type and quantity to obtain a target product positioning edge pixel point set;

[0034] Step S3.2: fitting a straight line to the target product positioning edge pixel point set to obtain a target product positioning edge straight line;

[0035] Step S3.3: determining a correction alignment transformation according to the target product positioning edge straight line and the standard product positioning edge straight line, and applying the correction alignment transformation to the target image to obtain a target correction alignment image.

[0036] According to the above technical solution, the specific steps of extracting the target 1D fingerprint from the target correction alignment image in step S4 are as follows:

[0037] Step S4.1: cutting a target template area image from the target correction alignment image;

[0038] Step S4.2: reducing the target template area image to obtain a target 1D fingerprint.

[0039] According to the above technical solution, the specific steps of determining the matching template information in the target correction alignment image according to the template group offset and the template information in the template group in step S6 are as follows:

[0040] Step S6.1: determining the position and size of each template in the template group aligned with the target according to the template group offset and the template information in the template group;

[0041] Step S6.2: listing the type, position and size of the template completely in the template area according to the position and size of each template in the template group aligned with the target.

[0042] The second aspect of the present application relates to a chip holder appearance defect detection template matching and positioning device, comprising a memory and one or more processors, the memory stores executable code, and the one or more processors execute the executable code to implement a chip holder appearance defect detection template matching and positioning method of the present application.

[0043] The third aspect of the present application relates to a computer readable storage medium having a program stored thereon, which is executed by a processor to implement a chip holder appearance defect detection template matching and positioning method of the present application.

[0044] The present application utilizes the shape characteristics of the chip holder, limits the template matching and positioning to one direction by positioning and aligning the product edge, limits the search range of the template matching and positioning to two product unit lengths by constructing a template group, and finally reduces the data calculation of the template matching and positioning from 2D to 1D by extracting and matching the 1D fingerprint, thereby finally realizing accurate and efficient chip holder appearance detection template matching and positioning. The chip holder appearance defect detection template matching and positioning method disclosed by the present application has three beneficial effects:

[0045] The present application searches for the product edge, corrects and aligns based on the product edge, limits the template matching and positioning to one direction (horizontal or vertical), reduces the search range, avoids the template matching and positioning errors or failures caused by product skew, and improves the accuracy and efficiency of the template matching and positioning;

[0046] The present application uses the shape characteristics of the chip support, designs and constructs a template group, limits the template matching and positioning search range to two product unit lengths, improves the efficiency of the template matching and positioning, and the matching and positioning of the template group composed of multiple templates have higher accuracy than the matching and positioning of a single template;

[0047] The present application extracts and matches 1D fingerprints, reduces the calculation amount by reducing the dimensionality of the template matching and positioning data from 2D to 1D, thereby improving the efficiency of the template matching and positioning.

[0048] Based on the above three aspects, the present application realizes accurate and efficient chip support appearance detection template matching and positioning. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 The present application defines the image detection area, template, template area and product unit distribution diagram for the template.

[0050] Figure 2 The present application defines the template group structure diagram. DETAILED DESCRIPTION

[0051] The present application discloses a chip support appearance defect detection template matching and positioning method, which is described in detail below in combination with preferred embodiments and with reference to the accompanying drawings.

[0052] Example 1

[0053] Referring to Figure 1 and Figure 2 , the chip support appearance defect detection template matching and positioning method of the present application comprises the following steps:

[0054] Step S1: defining the chip support template image, positioning the product edge and performing 2D Euclidean transformation on the image to obtain a corrected image;

[0055] Step S2: defining the template corrected image, constructing a template group and extracting a 1D fingerprint of the template group;

[0056] Step S3: defining the chip support detection target image, positioning the product edge and performing 2D Euclidean transformation on the image to obtain a target corrected alignment image;

[0057] Step S4: aligning the target image, and extracting a target 1D fingerprint;

[0058] Step S5: matching the template group 1D fingerprint with the target 1D fingerprint, and obtaining a template group offset;

[0059] Step S6: determining matched template information in the target alignment image according to the template group offset and template information in the template group.

[0060] The specific steps of the image positioning product edge and 2D Euclidean transformation of the image in step S1 based on the chip support template definition image are as follows:

[0061] Step S1.1: the template definition image is intercepted to obtain a detection area image, and the pixel values of the upper, lower, left and right four boundaries of the detection area image are respectively analyzed to determine the type and number of standard product positioning edges;

[0062] Step S1.2: for each standard product positioning edge, the product edge pixel points are searched from the boundary pixels of the detection area image outwardly and inwardly to obtain a standard product positioning edge pixel point set;

[0063] Step S1.3: for each standard product positioning edge pixel point set, a straight line is fitted to obtain a standard product positioning edge straight line;

[0064] Step S1.4: the standard product positioning edge straight line is used to determine a correction transformation, and the template definition image is subjected to the correction transformation to obtain a template definition correction image.

[0065] The specific steps of the template definition image interception in step S1.1 are as follows:

[0066] Step S1.1.1: the template definition image is intercepted to obtain a detection area image;

[0067] Step S1.1.2: the pixel values of the upper, lower, left and right four boundaries of the detection area image are respectively analyzed to obtain the mean value and standard deviation of the pixel values of the upper, lower, left and right four boundaries;

[0068] Step S1.1.3: according to the mean value and standard deviation of the pixel values of the upper, lower, left and right four boundaries of the detection area image, it is determined whether the product positioning edge is single or double, and which type of upper edge, lower edge, left edge or right edge each product positioning edge is.

[0069] One method of determining the type and number of product positioning edges in step S1.1.3 is to check the standard deviation of the pixel values of the image boundary of each detection area. If the standard deviation is less than a given threshold, then the boundary has product edges of the same type, otherwise the boundary has no corresponding product edges. For example, for the detection area image shown in FIG. 1 1, the left and right boundaries of the image are composed of background pixels, and the pixel values are substantially uniform. The standard deviation of the pixel values of the left boundary and the boundary pixels is very small, while the upper and lower boundaries have both foreground (product) pixels and background pixels, and the standard deviation of the pixel values will exceed the threshold. Thus, it is determined that there are two positioning edges, one on the left and one on the right. Figure 1

[0070] One method of searching for the product positioning edge pixel set in step S1.2 is to sample the boundary pixel set of the corresponding detection area (for example, selecting one pixel every five pixels) to obtain a boundary pixel set, and then traversing outward from each boundary pixel to check the pixel values. When a pixel value that is significantly different from the starting boundary pixel value is encountered (for example, the pixel value difference exceeds 10), the pixel is added to the product positioning edge pixel set.

[0071] The specific steps of applying the correction transformation to the template definition image to obtain the template definition correction image in step S1.4 are as follows:

[0072] Step S1.4.1: determining the rotation center point of the correction transformation according to the standard product positioning edge straight line;

[0073] Step S1.4.2: determining the rotation angle of the correction transformation according to the direction of the standard product positioning edge straight line;

[0074] Step S1.4.3: rotating the template definition image according to the rotation center point and the rotation angle of the correction transformation to obtain the template definition correction image.

[0075] One method of determining the rotation center point of the correction transformation according to the standard product positioning edge straight line in step S1.4.1 is as follows. When there is only one standard product positioning edge straight line, the midpoint of the line segment of the straight line within the detection area is selected as the rotation center point of the correction transformation. When there are two standard product positioning edge straight lines, the midpoints of the line segments of the two straight lines within the detection area are first determined, and then the midpoint of the line segment with the two midpoints as endpoints is selected as the rotation center point of the correction transformation.

[0076] As shown in FIG. 12, the specific steps of applying the correction transformation to the template definition image, constructing the template group, and extracting the 1D fingerprint of the template group in step S2 are as follows: Figure 2

[0077] ​​Step S2.1: 1D self-matching of the template definition rectified image to determine the product unit image size and the repeat pattern;

[0078] Step S2.2: extracting template images from the template definition rectified image according to the product unit image size and the repeat pattern, and splicing the template images according to the product unit repeat pattern to form a template group;

[0079] Step S2.3: reducing the spliced template group image to obtain a 1D fingerprint of the template group.

[0080] In particular, the method for determining the product unit image size in step S2.1 is to slide the template definition rectified image along the product positioning edge direction and compare and calculate the similarity of the overlapping part with the original image to construct a translation similarity vector, find the maximum value point in the similarity vector that exceeds a certain similarity threshold (for example, 0.95), and determine the product unit image size according to the average distance of the adjacent two maximum value points.

[0081] In particular, the method for determining the product unit repeat pattern in step S2.1 is to cut two adjacent product local images from the template definition rectified image along the product positioning edge direction according to the product unit image size, compare the two product local images, if there is no significant difference, the product unit has one type and the repeat pattern is AAAA, otherwise the product unit has two types and the repeat pattern is ABAB.

[0082] In particular, the method for extracting template images from the template definition rectified image according to the product unit image size and the repeat pattern in step S2.2 is that if the product unit repeat pattern is AAAA, one product unit rectangular frame is determined from the appropriate position of the template definition rectified image, then the size of the product unit rectangular frame is appropriately enlarged (for example, 20 pixels are added to the width and height), and one template image is cut according to the enlarged rectangular frame; if the product unit repeat pattern is ABAB, two adjacent product unit rectangular frames are determined from the appropriate position of the template definition rectified image, then the size of the product unit rectangular frame is appropriately enlarged, and finally two template images are cut according to the enlarged rectangular frame.

[0083] In particular, the method for splicing template images to form a template group according to the product unit repeat pattern in step S2.2 is to determine the number of templates n that can be completely accommodated in the detection area, and splice the corresponding template images of n+2 templates according to the product unit repeat pattern to form a template group. As shown in Figure 2 , the detection area can completely accommodate two templates, and the product unit repeat pattern is ABAB, then template A, template B, template A, and template B can be spliced in turn to form a template group Figure 2The order in the template B, template A, template B, template A, two ways are equivalent.

[0084] The method of reducing the spliced template group image in the step S2.3 to obtain the template group 1D fingerprint is that the template group image is reduced in the vertical direction of the positioning edge, the average value of the pixel value of each row or column of pixels is calculated, and the average value vector of the pixel value is taken as the template group 1D fingerprint.

[0085] The specific steps of the step S3 of locating the product edge and obtaining the target correction alignment image by performing 2D Euclidean transformation on the chip support detection target image are as follows:

[0086] Step S3.1: According to the type and number of the product positioning edge, the product edge pixel point set is obtained by searching the product edge pixel point from the boundary pixel of the detection area to the inside of the target product detection image.

[0087] Step S3.2: The straight line of the target product positioning edge pixel point set is fitted to obtain the target product positioning edge straight line.

[0088] Step S3.3: The correction alignment transformation is determined according to the target product positioning edge straight line and the standard product positioning edge straight line, and the target image is applied to the correction alignment transformation to obtain the target correction alignment image.

[0089] The specific steps of the step S4 of extracting the target 1D fingerprint from the target correction alignment image are as follows:

[0090] Step S4.1: The target template area image is cut from the target correction alignment image.

[0091] Step S4.2: The target 1D fingerprint is obtained by reducing the target template area image.

[0092] The specific steps of the step S6 of determining the matching template information in the target correction alignment image according to the template group offset and the template information in the template group are as follows:

[0093] Step S6.1: The position and size of each template in the template group aligned with the target are determined according to the template group offset and the template information in the template group.

[0094] Step S6.2: The type, position and size of the template completely in the template area are listed according to the position and size of each template in the template group aligned with the target.

[0095] Embodiment 2

[0096] The embodiment of the present application relates to a chip holder appearance defect detection template matching and positioning device, comprising a memory and one or more processors, the memory has executable code stored therein, and the one or more processors are used to implement a chip holder appearance defect detection template matching and positioning method of the embodiment 1 when the executable code is executed.

[0097] Embodiment 3

[0098] The embodiment of the present application relates to a computer readable storage medium, a program is stored on the computer readable storage medium, and the program is executed by a processor to implement a chip holder appearance defect detection template matching and positioning method of the embodiment 1.

[0099] The content described in the embodiments of the present application is only a list of implementation forms of the inventive concept, and the protection scope of the present application should not be regarded as limited to the specific forms stated in the embodiments, and the protection scope of the present application also extends to equivalent technical means that can be thought of by those skilled in the art according to the inventive concept.

Claims

1. A method for matching and locating templates for detecting appearance defects in chip supports, comprising the following steps: Step S1: Define an image for the chip carrier template, locate the product edge, and perform a 2D Euclidean transformation on the image to obtain a corrected image; Step S2: Define the corrected image for the template, construct the template group, and extract the 1D fingerprint of the template group; specifically including: Step S2.1: Perform 1D self-matching on the template definition correction image to determine the product unit image size and repetition pattern; Step S2.2: Extract template images from the template definition correction images according to the product unit image size and repetition pattern, and stitch the template images together according to the product unit repetition pattern to form a template group; Step S2.3: Reduce the spliced ​​template group images to obtain the template group 1D fingerprint; The method for determining the product unit image size in step S2.1 is to... The board defines a correction image that is translated and slid along the product positioning edge direction, and compares and calculates the similarity of the overlapping areas with the original image to construct a translation similarity vector. It then identifies the maxima exceeding a certain similarity threshold within the similarity vector and uses two adjacent maxima as the basis for further analysis. The average distance between points is used to determine the product unit image size. The method for determining the product unit repetition pattern in step S2.1 is to extract two adjacent product partial images from the template-defined correction image along the product positioning edge direction according to the product unit image size. For example, comparing two partial images of a product, if there are no significant differences, then the product unit has 1. If the product unit has two classes and the repetition pattern is AAAA, then the method for extracting the template image from the template definition correction image in step S2.2 based on the product unit image size and repetition pattern is as follows: If the product unit repetition pattern is AAAA, then determine one product unit rectangle from a suitable position in the template definition correction image, and extract one template image based on the enlarged rectangle; if the product unit has two classes and the repetition pattern is ABAB, then the method for extracting the template image from the template definition correction image is as follows: If the product unit repetition pattern is AAAA, then determine one product unit rectangle from a suitable position in the template definition correction image, and extract one template image based on the enlarged rectangle; if the product unit repetition pattern is AAAA ... determine one product unit rectangle from a suitable position in the template definition correction image, and extract one template image based on the enlarged rectangle; if the If the complex mode is ABAB, then determine two adjacent product unit rectangles from the appropriate position of the template-defined correction image, then keep the center position unchanged and appropriately enlarge the size of the product unit rectangles, and finally cut out two template images based on the enlarged rectangles. Step S3: Detect the target image of the chip bracket, locate the product edge, and perform a 2D Euclidean transformation on the image to obtain the target correction and alignment image; Step S4: Correct and align the target image, and extract the target 1D fingerprint. The method for extracting the 1D fingerprint is as follows: reduce the image according to the vertical direction of the positioning edge, and calculate each... The average pixel value of each row or column of pixels, and the vector of average pixel values ​​is used as a 1D representation. fingerprint; Step S5: Match the template group 1D fingerprint with the target 1D fingerprint to obtain the template group offset; Step S6: Determine the matching template information in the target correction and alignment image based on the template group offset and the template information in the template group.

2. The method for matching and locating templates for detecting appearance defects in chip carriers according to claim 1, characterized in that, Step S1 specifically includes: Step S1.1: Extract the detection area image from the template definition image, and perform statistical analysis on the pixel values ​​of the four boundaries of the detection area image (top, bottom, left, and right) to determine the standard product positioning edge type and quantity. Step S1.2: For each standard product positioning edge, search for product edge pixels from the outside to the inside starting from the boundary pixels of the detection area image to obtain the standard product positioning edge pixel set; Step S1.3: For each standard product positioning edge pixel set, fit a straight line to obtain the standard product positioning edge straight line; Step S1.4: Determine the correction transformation based on the standard product positioning edge line, apply the correction transformation to the template definition image, and obtain the template definition correction image.

3. The method for matching and locating templates for detecting appearance defects in chip carriers according to claim 2, characterized in that, Step S1.1 specifically includes: Step S1.1.1: Extract the detection region image from the template definition image; Step S1.1.2: Perform statistical analysis on the pixel values ​​of the four boundaries of the detection area image, and obtain the mean and standard deviation of the pixel values ​​of the four boundaries. Step S1.1.3: Based on the mean and standard deviation of the pixel values ​​of the four boundaries of the detection area image (top, bottom, left, and right), determine whether the product positioning edge is single-sided or double-sided, and which type of product positioning edge (top, bottom, left, or right) it is.

4. The method for matching and locating templates for detecting appearance defects in a chip carrier according to claim 2, characterized in that, Step S1.4 specifically includes: Step S1.4.1: Determine the center point of the correction transformation rotation based on the straight line of the standard product positioning edge; Step S1.4.2: Determine the correction and rotation angle based on the straight line direction of the positioning edge of the standard product; Step S1.4.3: Rotate the template definition image according to the rotation center point and rotation angle of the correction transformation to obtain the template definition correction image.

5. The method for matching and locating templates for detecting appearance defects in chip carriers according to claim 1, characterized in that, Step S3 specifically includes: Step S3.1: Based on the type and quantity of product positioning edges, search for product edge pixels in the chip bracket detection target image from the outer edge pixels to the inner edge pixels, starting from the boundary pixels of the detection area. Obtain the set of edge pixels for the target product; Step S3.2: Fit a straight line to the set of pixels on the target product's positioning edge to obtain the straight line for the target product's positioning edge; Step S3.3: Determine the correction alignment transformation based on the target product positioning edge line and the standard product positioning edge line, apply the correction alignment transformation to the target image, and obtain the target correction alignment image.

6. The method for matching and locating templates for detecting appearance defects in chip carriers according to claim 1, characterized in that, Step S4 specifically includes: Step S4.1: Extract the target template region image from the target correction and alignment image; Step S4.2: Reduce the target template region image to obtain the target 1D fingerprint.

7. The method for matching and locating templates for detecting appearance defects in chip carriers according to claim 1, characterized in that, Step S6 specifically includes: Step S6.1: Based on the template group offset and the template information in the template group, determine the position and size of each template in the template group that is aligned with the target; Step S6.2: Based on the position and size of each template in the template group aligned with the target, list the type, position, and size of the templates that are completely within the template area.

8. A template matching and positioning device for detecting appearance defects in chip brackets, characterized in that, The device includes a memory and one or more processors, wherein the memory stores executable code, and the one or more processors execute the executable code to implement a chip bracket appearance defect detection template matching and positioning method according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, It stores a program that, when executed by a processor, implements a chip bracket appearance defect detection template matching and positioning method according to any one of claims 1-7.

Images