Wafer thrust section classification detection method, system and device and storage medium
By combining the methods of object detection, image segmentation and image fitting, the problem of low accuracy and poor adaptability of wafer thrust cross-section classification detection is solved, and higher detection accuracy and fewer false detection are achieved.
Patent Information
- Application Number
- CN202510391080.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-31
- Publication Date
- 2025-07-11
AI Technical Summary
The accuracy of existing wafer thrust cross-section classification detection is not high, the adaptability is poor, and it is prone to mis-testing.
The method of combining object detection, image segmentation and image fitting is adopted to classify the categories of wafer thrust cross-sections through object detection technology, and pixel segmentation is used to divide the wafer area images of crystal residue categories, and circle fit the wafer sections obtained by pixel segmentation through image fitting technology. Finally, the area ratio calculation and threshold verification are performed based on the image characteristics of the wafer cross-section image and the cross-section fitting image.
It improves the accuracy and applicability of wafer thrust cross-section classification detection and reduces the situation of false detection.
Smart Images
Figure CN120298786A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of wafer packaging and testing, and more particularly, to a method, system, device, and storage medium for classifying and detecting wafer thrust cross-sections. Background Art
[0002] In the wafer-level packaging (WLP) industry, the wafer thrust cross-section is an important test step for evaluating wafer quality. The wafer thrust cross-section refers to the fracture surface formed by the separation and detachment of the chip from the packaging substrate; the purpose of wafer thrust cross-section detection is to evaluate the bonding strength between the chip and the packaging substrate, ensure that the chip does not fall off due to thermal stress, mechanical vibration, etc. during actual use, and can evaluate and optimize the reliability of the chip bonding process, which is the key to semiconductor packaging quality control.
[0003] Existing wafer thrust cross-section classification and detection technologies mainly use object detection models to classify and detect the bonding situation between the chip and the packaging substrate, directly determining the qualified (OK) and unqualified (NG) categories; however, due to the diverse bonding situations between the chip and the packaging substrate, the accuracy of directly classifying and detecting using existing wafer thrust cross-section classification and detection technologies is not high, and false detections are likely to occur, making it impossible to accurately evaluate and optimize the reliability of the chip packaging process. Summary of the Invention
[0004] The present invention provides a method, system, device, and storage medium for classifying and detecting wafer thrust cross-sections, which solves the problems of low accuracy, poor adaptability, and easy false detection in existing wafer thrust cross-section classification and detection.
[0005] In a first aspect, an embodiment of the present invention provides a method for classifying and detecting wafer thrust cross-sections, the method including the following processes:
[0006] Input a wafer packaging image into an object detection model for classification and detection to obtain the category and position of the wafer thrust cross-section; wherein, the category of the wafer thrust cross-section includes qualified, unqualified, and crystal residue.
[0007] If the category of the wafer thrust cross-section is crystal residue, then intercept the wafer packaging image based on the position of the wafer thrust cross-section to obtain a wafer region image.
[0008] Perform pixel segmentation on the wafer region image to obtain a wafer cross-section image, and obtain the normal cross-section area based on the wafer cross-section image.
[0009] Extract the edge contour and perform image fitting on the wafer region image to obtain a wafer cross-section fitting image; and obtain the fitting circle area of the cross-section based on the wafer cross-section fitting image.
[0010] Calculate the area ratio of the normal area of the cross-section and the area of the fitted circle of the cross-section, and classify and detect the crystal residue category based on the area ratio to obtain the classification and detection result; wherein, the classification and detection result includes qualified and unqualified.
[0011] In the above embodiments, the present invention realizes the classification and detection of the wafer thrust cross-section by combining object detection, image segmentation and image fitting. First, the object detection technology is used to classify the category of the wafer thrust cross-section, then the image segmentation technology is used to perform pixel segmentation on the wafer area image of the crystal residue category, and the image fitting technology is used to perform circle fitting on the wafer cross-section obtained by pixel segmentation. Finally, the area ratio calculation and threshold verification are performed by combining the image characteristics of the wafer cross-section image and the cross-section fitting image, thereby accurately classifying and detecting the wafer thrust cross-section, and solving the problems of low accuracy, poor adaptability and easy misdetection in the existing classification and detection of the wafer thrust cross-section.
[0012] As some alternative embodiments of the present application, the process of edge contour extraction and image fitting for the wafer area image is as follows:
[0013] Perform edge contour extraction on the wafer area image to obtain the edge coordinates of the wafer cross-section;
[0014] Use the least squares method to perform image fitting on the edge coordinates of the wafer cross-section to obtain the fitted image of the wafer cross-section.
[0015] In the above embodiments, the present invention combines the edge contour extraction technology and the image fitting technology, and can quickly obtain the fitted image of the wafer cross-section.
[0016] As some alternative embodiments of the present application, the process of using the least squares method to perform image fitting on the edge coordinates of the wafer cross-section is as follows:
[0017] Perform least squares method processing on the edge coordinates of the wafer cross-section to obtain the radius of the fitted circle and the center coordinates;
[0018] Based on the radius of the fitted circle and the center coordinates, perform image fitting to obtain the fitted image of the wafer cross-section.
[0019] In the above embodiments, the present invention performs circle fitting using the least squares method for the geometric characteristics of the wafer cross-section, and can accurately obtain the fitted circle of the wafer cross-section.
[0020] As some alternative embodiments of the present application, the process of performing least squares method processing on the edge coordinates of the wafer cross-section is as follows:
[0021] Assume that the radius of the fitted circle is R and the center coordinates are (A, B), then let the coefficient a = -2A, b = -2B, c = A 2 +B2 -R 2 ;
[0022] Assume that the edge coordinates of the wafer cross-section correspond to N measurement points (X i , Y i ), where i = 1, 2,..., N. Then the square of the distance from the i-th measurement point to the center of the circle is d i 2 =(X i -A) 2 +(Y i -B) 2 , where the error from the measurement point to the fitted circle is δ i =d i 2 -R 2 =X i 2 +Y i 2 +aX i +bY i +c. The sum of the squared errors of all measurement points is
[0023] Minimize and take partial derivatives of the squared error Q(a, b, c) of all measurement points to obtain a set of linear equations, and solve the set of linear equations to obtain the values of coefficients a, b, and c;
[0024] Calculate the center coordinates A = -0.5a, B = -0.5b, and the radius
[0025] As some alternative embodiments of the present application, the process for classifying and detecting crystal residue categories based on the area ratio is as follows:
[0026] Perform threshold detection on the area ratio of the normal cross-section area and the fitted circle area of the cross-section;
[0027] If the area ratio is greater than the threshold, the original crystal residue category of the wafer thrust cross-section is changed to qualified; otherwise, the original crystal residue category of the wafer thrust cross-section is changed to unqualified.
[0028] In the above embodiments, the present invention classifies the wafer cross-sections with specific rules into crystal residue categories, and fits a circle through the pixel segmentation result to calculate a quantifiable index. If the rules change later, only the threshold needs to be changed, without retraining the model, and the classification detection has higher accuracy and stronger applicability.
[0029] In some alternative embodiments of the present application, before inputting the wafer package image into the object detection model for classification detection, it is necessary to perform iterative training on the object detection model.
[0030] In some alternative embodiments of the present application, the process of iterative training of the object detection model is as follows:
[0031] Label the category and position of the wafer thrust section on the surface of the wafer package image to obtain a target box containing the category and position of the wafer thrust section;
[0032] Input the wafer package image containing the target box into the object detection model for iterative training to obtain the iteratively trained object detection model.
[0033] In a second aspect, the present invention provides a wafer thrust section classification detection system, the system includes:
[0034] An object classification unit, which is used to input the wafer package image into the object detection model for classification detection to obtain the category and position of the wafer thrust section; wherein, the categories of the wafer thrust section include qualified, unqualified, and crystal residue;
[0035] An image cropping unit, which is used to judge whether the category of the wafer thrust section is crystal residue. If the category of the wafer thrust section is crystal residue, the wafer package image is cropped based on the position of the wafer thrust section to obtain a wafer area image;
[0036] A pixel segmentation unit, which is used to perform pixel segmentation on the wafer area image to obtain a wafer cross-section image, and obtain the normal cross-section area based on the wafer cross-section image;
[0037] An image fitting unit, which is used to extract the edge contour and perform image fitting on the wafer area image to obtain a wafer cross-section fitting image; and obtain the fitting circle area of the cross-section based on the wafer cross-section fitting image;
[0038] An optimization classification unit, which is used to calculate the area ratio of the normal cross-section area and the fitting circle area of the cross-section, and perform classification detection on the crystal residue category based on the area ratio to obtain a classification detection result; wherein, the classification detection result includes qualified and unqualified.
[0039] In a third aspect, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned wafer thrust section classification detection method.
[0040] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for classifying and detecting a wafer thrust cross-section is implemented.
[0041] The beneficial effects of the present invention are as follows: The present invention combines object detection, image segmentation, and image fitting to perform classification and detection of wafer thrust cross-sections. First, object detection technology is used to classify the categories of wafer thrust cross-sections. Then, image segmentation technology is used to perform pixel segmentation on the wafer area image of the crystal residue category, and image fitting technology is used to perform circle fitting on the wafer cross-section obtained by pixel segmentation. Finally, the area ratio calculation and threshold verification are performed by combining the image characteristics of the wafer cross-section image and the cross-section fitting image, which can accurately classify and detect the wafer thrust cross-section, has strong applicability, and is not prone to false detection. Description of the Drawings
[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.
[0043] Figure 1 It is a schematic structural diagram of a computer device in the hardware operating environment of the embodiments of the present invention;
[0044] Figure 2 It is a flowchart of the method for classifying and detecting a wafer thrust cross-section according to the embodiments of the present invention;
[0045] Figure 3 It is a schematic diagram of three classification results of wafer thrust cross-sections output by the object detection model according to the embodiments of the present invention;
[0046] Figure 4 It is a schematic diagram of the normal area of the cross-section according to the embodiments of the present invention;
[0047] Figure 5 It is a schematic diagram of the area of the cross-section fitting circle and the area of the intersecting cross-section fitting circle according to the embodiments of the present invention. Detailed Embodiments
[0048] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0049] To solve the problems of low accuracy and poor applicability in the existing classification detection of wafer thrust cross-sections, and prone to false detection. This application provides a method, system, device, and storage medium for classifying and detecting wafer thrust cross-sections. Before introducing the specific technical solutions of this application, the hardware operating environment involved in the embodiment solutions of this application will be introduced first.
[0050] Please refer to Figure 1 , Figure 1 which is a schematic structural diagram of a computer device for the hardware operating environment involved in the embodiment solutions of this application.
[0051] As Figure 1 shown, the computer device may include: a processor, such as a Central Processing Unit (CPU), a communication bus, a user interface, a network interface, and a memory. Among them, the communication bus is used to realize the connection and communication between these components. The user interface may include a display screen (Display) and an input unit such as a keyboard (Keyboard). Optionally, the user interface may further include a standard wired interface and a wireless interface. The network interface may optionally include a standard wired interface and a wireless interface (such as a Wi-Fi interface). The memory may be a high-speed random access memory (Random Access Memory, RAM) or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. Optionally, the memory may also be a storage device independent of the aforementioned processor.
[0052] Those skilled in the art can understand that Figure 1 the structure shown in
[0053] does not constitute a limitation on the computer device, and may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements. Figure 1 As
[0054] shown, the memory as a storage medium may include an operating system, a network communication module, a user interface module, and an electronic program module. Figure 1 In the computer device shown in
[0055] Based on the hardware environment of the foregoing embodiments, the embodiments of this application provide a method for classifying and detecting wafer thrust cross-sections. Please refer to Figure 2 ,Figure 2 It is a flowchart of the classification detection method for the wafer thrust cross-section, and the method process is as follows:
[0056] (1) Input the wafer package image into the target detection model for classification detection to obtain the category and position of the wafer thrust cross-section; wherein, the categories of the wafer thrust cross-section include qualified, unqualified, and crystal residue.
[0057] In an embodiment of the present invention, before inputting the wafer package image into the target detection model for classification detection, it is necessary to perform iterative training on the target detection model; the target detection model includes but is not limited to YOLO, ResNet, MobileNet, etc.
[0058] Specifically, the process of iterative training of the target detection model is as follows:
[0059] (1.1) Label the category and position of the wafer thrust cross-section on the surface of the wafer package image to obtain a target box containing the category and position of the wafer thrust cross-section.
[0060] (1.2) Input the wafer package image containing the target box into the target detection model for iterative training to obtain the iteratively trained target detection model.
[0061] In an embodiment of the present invention, through iterative training of the target detection model, the target detection model after iterative training classifies the wafer thrust cross-section that completely covers the through-hole surface of the package substrate (or directly covers the package substrate surface) as qualified (OK), classifies the wafer thrust cross-section that covers the through-hole of the package substrate as unqualified (NG), and classifies the wafer thrust cross-section with specific rules remaining near the through-hole of the package substrate as crystal residue (TEMP). Please refer to Figure 3 , Figure 3 which is a schematic diagram of the three classification results of the wafer thrust cross-section output by the target detection model.
[0062] (2) Determine whether the category of the wafer thrust cross-section is crystal residue. If the category of the wafer thrust cross-section is crystal residue, then intercept the wafer package image based on the position of the wafer thrust cross-section to obtain a wafer area image.
[0063] (3) Perform pixel segmentation on the wafer area image to obtain a wafer cross-section image, and obtain the normal cross-section area based on the wafer cross-section image. Specifically, first automatically calculate the number of pixels through image scanning, then calculate the area = number of pixels × actual area represented by a single pixel, and further obtain the normal cross-section area Area. Please refer to Figure 4 , Figure 4 which is a schematic diagram of the normal cross-section area.
[0064] (4) Perform edge contour extraction and image fitting on the wafer area image to obtain the wafer cross-section fitting image and the via fitting image; and obtain the cross-section fitting circle area and the intersecting cross-section fitting circle area based on the wafer cross-section fitting image and the via fitting image.
[0065] In the embodiment of the present invention, the process of performing edge contour extraction and image fitting on the wafer area image and the via image is as follows:
[0066] (4.1) Perform edge contour extraction on the wafer area image and the via image to obtain the edge coordinates of the wafer cross-section and the edge coordinates of the via.
[0067] (4.2) Use the least squares method to perform image fitting on the edge coordinates of the wafer cross-section and the edge coordinates of the via to obtain the cross-section fitting image and the via fitting image.
[0068] In the embodiment of the present invention, the process of using the least squares method to perform image fitting on the edge coordinates of the wafer cross-section and the edge coordinates of the via is as follows:
[0069] (4.21) Perform least squares method processing on the edge coordinates of the wafer cross-section and the edge coordinates of the via to obtain the radii and center coordinates of two fitting circles.
[0070] Specifically, the process of performing least squares method processing on the edge coordinates of the wafer cross-section and the edge coordinates of the via is as follows:
[0071] ① Determine the undetermined coefficients;
[0072] Assume that the radius of the fitting circle is R and the center coordinates are (A, B);
[0073] The equation of the circle can be expressed as: (x - A) 2 +(y - B) 2 =R 2 , expand the equation to get: x 2 +y 2 -2Ax - 2By + A 2 +B 2 -R 2 =0;
[0074] To simplify the calculation, let the coefficients a = -2A, b = -2B, c = A 2 +B 2 -R 2 , then the equation of the circle can be expressed as: x 2 +y 2 +ax + by + c = 0;
[0075] Therefore, the parameters to be solved are a, b, c.
[0076] ② Calculate the sum of squared errors;
[0077] Assume that the edge coordinates of the wafer cross-section correspond to N measurement points (X i , Y i ), where i = 1, 2,..., N;
[0078] The square of the distance from the i-th measurement point to the center of the circle is: d i 2 = (X i - A) 2 + (Y i - B) 2 ;
[0079] The error from the measurement point to the fitted circle is: δ i = d i 2 - R 2 = X i 2 + Y i 2 + aX i + bY i + c;
[0080] The sum of squared errors of all measurement points is:
[0081] ③ Solve for the minimum sum of squared errors;
[0082] Minimize and take the partial derivatives of the sum of squared errors Q(a, b, c) of all measurement points to obtain a set of linear equations;
[0083] Obtain the values of coefficients a, b, and c by solving the linear equations.
[0084] ④ Calculate the center of the circle and the radius;
[0085] Calculate the center coordinates of the circle A = -0.5a, B = -0.5b, and the radius
[0086] Through the above least squares method, the center coordinates (A1, B1) and radius R1 of the cross-section fitted circle, and the center coordinates (A2, B2) and radius R2 of the via-fitted circle can be obtained.
[0087] (4.22) Perform image fitting based on the radius and center coordinates of the fitted circle to obtain the cross-section fitted image of the wafer and the via-fitted image.
[0088] Specifically, the calculation formulas for the area of the cross-section fitted circle and the area of the intersecting cross-section fitted circle are:
[0089] Cross-sectional fitting circle area: Area1 = πR1 2 ;
[0090] Intersecting cross-sectional fitting circle area: Area2 = πR2 2 - Area0, where Area0 represents the intersecting area of the two fitting circles;
[0091] In the embodiments of the present invention, please refer to Figure 5 , Figure 5 which is a schematic diagram of the cross-sectional fitting circle area and the intersecting cross-sectional fitting circle area. The calculation formula for the intersecting area of the two fitting circles is as follows:
[0092] ① Calculate the center distance d between the cross-sectional fitting circle area and the center of the through-hole fitting circle.
[0093] ② Calculate the central angle of the intersecting segment area between the wafer cross-sectional fitting circle and the through-hole fitting circle.
[0094] For the wafer cross-sectional fitting circle with radius R1, the central angle is:
[0095]
[0096] For the through-hole fitting circle with radius R2, the central angle is:
[0097]
[0098] ③ Calculate the intersecting segment area.
[0099] For the segment area with radius R1: A1 = 0.5R1 2 (θ1 - sinθ1);
[0100] For the segment area with radius R2: A2 = 0.5R2 2 (θ2 - sinθ2);
[0101] The common intersecting area: Area0 = A1 + A2;
[0102] (5) Calculate the area ratio of the cross-sectional normal area and the cross-sectional fitting circle area, and classify and detect the crystal residue category based on the area ratio to obtain the classification and detection result.
[0103] In the embodiments of the present invention, the process of classifying and detecting the crystal residue category based on the area ratio is as follows:
[0104] (5.1) Perform a threshold detection on the area ratio of the cross-sectional normal area and the cross-sectional fitting circle area.
[0105] (5.2) If the area ratio is greater than the threshold, the original crystal residue category of the wafer thrust section is changed to qualified; otherwise, the original crystal residue category of the wafer thrust section is changed to unqualified.
[0106] For example, if the area ratio K = Area / Area1 is greater than 80%, the original crystal residue category of the wafer thrust section is changed to qualified; if the area ratio is less than or equal to 80%, the original crystal residue category of the wafer thrust section is changed to unqualified.
[0107] Meanwhile, when the fitting circle of the wafer cross-section intersects with the adjacent through-holes, that is, when a part of the wafer bonding cross-section is located in the through-holes of the packaging substrate, it will affect the classification inspection of the wafer thrust section. Therefore, while classifying and detecting the crystal residue category based on the area ratio, the intersection part deduction check method can be used to verify the results of the classification detection.
[0108] (6) Obtain the area of the fitting circle of the intersection cross-section, calculate the area ratio of the fitting circle of the intersection cross-section and the fitting circle of the cross-section, and verify the classification result of the crystal residue category based on the area ratio to obtain the final classification detection result.
[0109] In the embodiment of the present invention, the process of verifying the classification result of the crystal residue category based on the area ratio is as follows:
[0110] (6.1) Perform threshold detection on the area ratio of the fitting circle of the intersection cross-section and the fitting circle of the cross-section.
[0111] (6.2) If the area ratio is greater than the threshold and the original crystal residue category of the wafer thrust section is changed to qualified, then keep the category of the wafer thrust section as qualified; otherwise, the original crystal residue category of the wafer thrust section is changed to unqualified.
[0112] For example, if the area ratio k1 = Area / Area2 is greater than 70% and the area ratio K = Area / Area1 is greater than 80%, the original crystal residue category of the wafer thrust section is changed to qualified; otherwise, the original crystal residue category of the wafer thrust section is changed to unqualified.
[0113] In summary, the present invention realizes the classification detection of the wafer thrust cross-section by combining object detection, image segmentation, and image fitting. First, object detection technology is used to classify the category of the wafer thrust cross-section. Then, image segmentation technology is used to perform pixel segmentation on the wafer area image of the crystal residue category, and image fitting technology is used to perform circle fitting on the wafer cross-section obtained by pixel segmentation. Finally, the area ratio calculation and threshold verification are performed by combining the image characteristics of the wafer cross-section image and the cross-section fitting image, thereby accurately classifying and detecting the wafer thrust cross-section, and solving the problems of low accuracy, poor adaptability, and easy misdetection in the existing classification detection of the wafer thrust cross-section.
[0114] In addition, in one embodiment, based on the same inventive concept as the foregoing embodiment, the embodiment of the present invention provides a wafer thrust cross-section classification detection system, which corresponds one-to-one with the method of the foregoing embodiment 1. The system includes:
[0115] An object classification unit, which is configured to input a wafer package image into an object detection model for classification detection to obtain the category and position of the wafer thrust cross-section; wherein, the category of the wafer thrust cross-section includes qualified, unqualified, and crystal residue;
[0116] An image intercepting unit, which is configured to determine whether the category of the wafer thrust cross-section is crystal residue. If the category of the wafer thrust cross-section is crystal residue, the wafer package image is intercepted based on the position of the wafer thrust cross-section to obtain a wafer area image;
[0117] A pixel segmentation unit, which is configured to perform pixel segmentation on the wafer area image to obtain a wafer cross-section image, and obtain a normal cross-section area based on the wafer cross-section image;
[0118] An image fitting unit, which is configured to extract the edge contour and perform image fitting on the wafer area image to obtain a wafer cross-section fitting image; and obtain a cross-section fitting circle area based on the wafer cross-section fitting image;
[0119] An optimization classification unit, which is configured to calculate the area ratio of the normal cross-section area and the cross-section fitting circle area, and perform classification detection on the crystal residue category based on the area ratio to obtain a classification detection result; wherein, the classification detection result includes qualified and unqualified.
[0120] It should be noted that each unit in the wafer thrust cross-section classification detection system in this embodiment corresponds one-to-one with each step in the wafer thrust cross-section classification detection method in the foregoing embodiment. Therefore, the specific implementation manner and the achieved technical effects of this embodiment can refer to the implementation manner of the foregoing wafer thrust cross-section classification detection method, which will not be elaborated here.
[0121] In addition, in one embodiment, the present application further provides a computer device, which includes a processor, a memory, and a computer program stored in the memory. When the computer program is run by the processor, it implements the method in the foregoing embodiment.
[0122] In addition, in one embodiment, the present application further provides a computer storage medium, on which a computer program is stored. When the computer program is run by the processor, it implements the method in the foregoing embodiment.
[0123] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disc, or CD-ROM; or it may be various devices including one or any combination of the foregoing memories. The computer may be various computing devices including smart terminals and servers.
[0124] In some embodiments, the executable instructions may be in the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including being deployed as an independent program or being deployed as a module, component, subroutine, or other unit suitable for use in a computing environment.
[0125] As an example, the executable instructions may or may not correspond to a file in the file system, and may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a HyperText Markup Language (HTML) document, stored in a single file dedicated to the program being discussed, or, stored in multiple cooperating files (for example, files that store one or more modules, subroutines, or portions of code).
[0126] As an example, the executable instructions may be deployed to execute on one computing device, or on multiple computing devices located at one location, or, on multiple computing devices distributed at multiple locations and interconnected by a communication network.
[0127] It should be noted that, in this document, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article or system comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or system. Without further limitation, an element defined by the statement "comprising one..." does not exclude the existence of additional identical elements in the process, method, article or system comprising such element.
[0128] The serial numbers of the embodiments of the present application above are only for description and do not represent the superiority or inferiority of the embodiments.
[0129] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions for causing a multimedia terminal device (which can be a mobile phone, computer, television receiver, or network device, etc.) to execute the methods described in the various embodiments of the present application.
[0130] The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present application.
Claims
1. A method for classifying and detecting a wafer thrust cross-section, characterized in that, The method includes the following processes: Input the wafer package image into the target detection model for classification detection to obtain the category and position of the wafer thrust section; wherein, the categories of the wafer thrust section include qualified, unqualified, and crystal residue; If the category of the wafer thrust section is crystal residue, intercept the wafer package image based on the position of the wafer thrust section to obtain the wafer area image; Perform pixel segmentation on the wafer area image to obtain the wafer cross-section image, and obtain the normal cross-section area based on the wafer cross-section image; Extract the edge contour and perform image fitting on the wafer area image to obtain the wafer cross-section fitting image; and obtain the fitting circle area of the cross-section based on the wafer cross-section fitting image; Calculate the area ratio of the normal cross-section area and the fitting circle area of the cross-section, and perform classification detection on the crystal residue category based on the area ratio to obtain the classification detection result; wherein, the classification detection result includes qualified and unqualified.
2. The method for classifying and detecting the wafer thrust cross-section according to claim 1, characterized in that, The process of extracting the edge contour and performing image fitting on the wafer area image is as follows: Extract the edge contour of the wafer area image to obtain the edge coordinates of the wafer cross-section; Use the least squares method to perform image fitting on the edge coordinates of the wafer cross-section to obtain the wafer cross-section fitting image.
3. A method for classifying and detecting the wafer thrust cross-section according to claim 2, characterized in that, The process of using the least squares method to perform image fitting on the edge coordinates of the wafer cross-section is as follows: Perform least squares method processing on the edge coordinates of the wafer cross-section to obtain the radius and center coordinates of the fitting circle; Perform image fitting based on the radius and center coordinates of the fitting circle to obtain the wafer cross-section fitting image.
4. The method for classifying and detecting the wafer thrust cross-section according to claim 3, characterized in that, The process of performing least squares method processing on the edge coordinates of the wafer cross-section is as follows: Assume that the radius of the fitted circle is R and the center coordinates are (A, B), then let the coefficient a = -2A, b = -2B, and c = A 2 +B 2 -R 2 ; Assume that the edge coordinates of the wafer cross-section correspond to N measurement points (X i , Y i ), where i = 1, 2,..., N. Then the square of the distance from the i-th measurement point to the center of the circle is d i 2 = (X i - A) 2 + (Y i - B) 2 . Among them, the error between the measurement point and the fitted circle is δ i = d i 2 - R 2 = X i 2 + Y i 2 + aX i + bY i + c. The sum of the squares of the errors of all measurement points is Minimize and take partial derivatives of the error square Q(a, b, c) of all measurement points to obtain a set of linear equations, and solve the linear equations to obtain the values of coefficients a, b, and c; Calculate the center coordinates A = -0.5a, B = -0.5b and the radius according to the values of coefficients a, b, and c 5. A method for classifying and detecting a wafer thrust cross-section according to claim 1, characterized in that The process of performing classification detection on the crystal residue category based on the area ratio is as follows: Perform threshold detection on the area ratio of the normal cross-section area and the fitting circle area of the cross-section; If the area ratio is greater than the threshold, change the original crystal residue category of the wafer thrust section to qualified, otherwise, change the original crystal residue category of the wafer thrust section to unqualified.
6. A method for classifying and detecting a wafer thrust cross-section according to claim 1, characterized in that, Before inputting the wafer package image into the target detection model for classification detection, the target detection model needs to be iteratively trained.
7. A wafer thrust cross-section classification detection method according to claim 6, characterized in that The process of iterative training of the target detection model is as follows: Label the category and position of the wafer thrust section corresponding to the surface of the wafer package image to obtain a target box containing the category and position of the wafer thrust section; Input the wafer package image containing the target box into the target detection model for iterative training to obtain the iteratively trained target detection model.
8. A wafer thrust cross-section classification detection system, characterized in that, The system includes: A target classification unit, which is used to input the wafer package image into the target detection model for classification detection to obtain the category and position of the wafer thrust section; wherein, the categories of the wafer thrust section include qualified, unqualified, and crystal residue; An image capture unit, which is used to determine whether the category of the wafer thrust cross-section is crystal residue. If the category of the wafer thrust cross-section is crystal residue, the wafer packaging image is captured based on the position of the wafer thrust cross-section to obtain a wafer area image; A pixel segmentation unit, which is used to perform pixel segmentation on the wafer area image to obtain a wafer cross-section image, and obtain the normal cross-section area based on the wafer cross-section image; An image fitting unit, which is used to extract the edge contour and perform image fitting on the wafer area image to obtain a wafer cross-section fitting image; and obtain the fitting circle area of the cross-section based on the wafer cross-section fitting image; An optimization classification unit, which is used to calculate the area ratio of the normal cross-section area and the fitting circle area of the cross-section, and perform classification detection on the crystal residue category based on the area ratio to obtain a classification detection result; wherein, the classification detection result includes qualified and unqualified.
9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the method for classifying and detecting a wafer thrust cross-section according to any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by the processor, it implements the method for classifying and detecting a wafer thrust cross-section according to any one of claims 1-7.