CN-121998933-A - Wafer image processing method, electronic equipment and storage medium

CN121998933ACN 121998933 ACN121998933 ACN 121998933ACN-121998933-A

Abstract

The application provides a wafer image processing method, electronic equipment and a storage medium, wherein the method comprises the steps of obtaining multi-frame original images of a wafer to be detected and multi-mode scanning information bound with each original image, wherein the multi-mode scanning information comprises Z-axis height, illumination mode and two-dimensional coordinates of a two-dimensional platform where the wafer to be detected is located when the original images are shot; and according to the multi-mode scanning information, splicing the processed images to obtain spliced images, wherein the spliced images are used as input images of a defect recognition model, so that the defect recognition model detects defects of a wafer to be detected based on the spliced images, and the processing effect of the wafer images is improved.

Inventors

ZHANG QI
FU YOUYIN
LIU XUESONG

Assignees

新毅东(北京)科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260121

Claims (10)

1. A wafer image processing method, comprising: Acquiring multi-frame original images of a wafer to be tested and multi-mode scanning information bound with each original image, wherein the multi-mode scanning information comprises a Z-axis height, an illumination mode and a two-dimensional coordinate of a two-dimensional platform where the wafer to be tested is located when the original images are shot; Denoising each original image according to the multi-mode scanning information to obtain each processed image; and performing stitching processing on each processed image according to the multi-mode scanning information to obtain a stitched image, wherein the stitched image is used as an input image of a defect recognition model, so that the defect recognition model performs defect detection on the wafer to be detected based on the stitched image.
2. The method according to claim 1, wherein the denoising processing is performed on each of the original images according to the multi-mode scanning information, so as to obtain each processed image, and the method comprises: determining a target denoising processing algorithm according to the illumination mode; and denoising the original image by adopting the target denoising algorithm according to the multi-mode scanning information to obtain a processed image.
3. The method of claim 2, wherein said determining a target denoising process algorithm from the illumination pattern comprises: If the illumination mode is a bright field illumination mode, determining the target denoising processing algorithm as a guiding filtering algorithm introducing Z-axis height; If the illumination mode is a dark field illumination mode, determining that the target denoising processing algorithm is a limiting contrast adaptive histogram equalization (CLAHE) algorithm.
4. The method of claim 3, wherein denoising the original image using the target denoising algorithm according to the multi-modal scan information to obtain a processed image, comprising: Acquiring a template image of the wafer to be tested in the bright field illumination mode; And according to the template image and the Z-axis height when the original image is shot, adopting the guide filtering algorithm for introducing the Z-axis height to perform linear filtering processing on the original image, and obtaining the processed image.
5. The method according to claim 4, wherein the performing linear filtering processing on the original image by using the guided filtering algorithm for introducing the Z-axis height according to the template image and the Z-axis height when the original image is captured comprises: determining a guide image of the original image according to the template image; Determining the defocus height according to the Z-axis height and a preset target focal plane height; In a local window corresponding to the original image, determining a minimized guide filtering cost function corresponding to the local window according to the guide image, the defocus height and the original image by taking the error minimization of the processed image and the original image in the local window as a target; And determining a first target linear coefficient and a second target linear coefficient corresponding to the local window according to the minimized guided filtering cost function corresponding to the local window, and performing linear filtering processing on the original image according to the first target linear coefficient, the second target linear coefficient and the guided image.
6. The method of claim 3, wherein denoising the original image using the target denoising algorithm according to the multi-modality scan information to obtain a processed image, further comprising: In the dark field illumination mode, dividing the original image into a plurality of subareas by adopting a CLAHE algorithm, and respectively carrying out histogram equalization treatment on each subarea to obtain a histogram of each subarea; Cutting and reassigning the histogram of each subarea according to a preset limiting threshold value to obtain a processed histogram of each subarea; And determining a cumulative distribution function CDF of the original image according to the processed histogram of each subarea, and carrying out equalization processing on the original image according to the CDF to obtain the processed image.
7. The method according to claim 1, wherein the performing a stitching process on each of the processed images according to the multi-modal scan information to obtain a stitched image includes: Performing preliminary stitching processing on each processed image according to the two-dimensional coordinates of a two-dimensional platform where the wafer to be tested is located when each original image is shot, so as to obtain a preliminary stitched image; And according to the two-dimensional coordinates of a two-dimensional platform where the wafer to be detected is positioned when each original image is shot, adopting a sub-pixel registration algorithm to adjust the overlapping area of each adjacent processed image in the preliminary spliced image, and obtaining the spliced image.
8. The method according to claim 7, wherein the adjusting the overlapping area of each adjacent processed image in the preliminary stitched image by using a subpixel registration algorithm according to the two-dimensional coordinates of the two-dimensional platform on which the wafer to be measured is located when each original image is captured comprises: determining whether an overlapping region exists between the first processed image and an adjacent second processed image; If yes, extracting an overlapping area from the first processed image to serve as a first image, and extracting an overlapping area from the second processed image to serve as a second image; performing Fourier transform and inverse Fourier transform on the first image and the second image respectively by adopting the sub-pixel registration algorithm to obtain sub-pixel level displacement deviation of the first processed image and the second processed image; And adjusting the position of the second processed image according to the sub-pixel level displacement deviation to obtain an adjusted second processed image, so that an overlapping area does not exist between the first processed image and the adjacent adjusted second processed image.
9. An electronic device comprising a processor, a memory and a bus, the memory storing machine readable instructions executable by the processor, the processor and the memory in communication via the bus when the electronic device is in operation, the processor executing the machine readable instructions to perform the steps of the image processing method of any of claims 1 to 8.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, performs the steps of the image processing method according to any one of claims 1 to 8.

Description

Wafer image processing method, electronic equipment and storage medium Technical Field The present application relates to the field of image processing technologies, and in particular, to a wafer image processing method, an electronic device, and a storage medium. Background In the semiconductor manufacturing process, wafer defect detection is a key link for guaranteeing the chip yield. The surface of the wafer is usually scanned and imaged by a high-resolution camera, and the accuracy of the subsequent defect identification is improved through image preprocessing steps such as noise reduction, enhancement, splicing and the like. However, in the prior art, the image preprocessing modules operate independently, so that the preprocessing strategy lacks pertinence, and the contradiction between noise suppression and weak defect retention is difficult to be considered. For example, common noise reduction algorithms may smooth out weak scratch defects, while excessive enhancement amplifies noise, creating false defects. In addition, the prior art is mostly dependent on a matching algorithm based on characteristic points during image splicing, fuzzy characteristic points are easy to generate, and splicing dislocation or ghosting is caused. Therefore, in the prior art, there is a limitation in preprocessing the wafer image during the wafer defect detection process. Disclosure of Invention The application aims to overcome the defects in the prior art and provide a wafer image processing method, electronic equipment and a storage medium, so as to solve the problem of the prior art that the preprocessing of wafer images in the wafer defect detection process has a certain limitation. In order to achieve the above purpose, the technical scheme adopted by the embodiment of the application is as follows: In a first aspect, an embodiment of the present application provides a wafer image processing method, including: Acquiring multi-frame original images of a wafer to be tested and multi-mode scanning information bound with each original image, wherein the multi-mode scanning information comprises a Z-axis height, an illumination mode and a two-dimensional coordinate of a two-dimensional platform where the wafer to be tested is located when the original images are shot; Denoising each original image according to the multi-mode scanning information to obtain each processed image; and performing stitching processing on each processed image according to the multi-mode scanning information to obtain a stitched image, wherein the stitched image is used as an input image of a defect recognition model, so that the defect recognition model performs defect detection on the wafer to be detected based on the stitched image. As an optional implementation manner, the denoising processing is performed on each original image according to the multi-mode scanning information, so as to obtain each processed image, and the method includes: determining a target denoising processing algorithm according to the illumination mode; and denoising the original image by adopting the target denoising algorithm according to the multi-mode scanning information to obtain a processed image. As an optional implementation manner, the determining a target denoising processing algorithm according to the illumination mode includes: If the illumination mode is a bright field illumination mode, determining the target denoising processing algorithm as a guiding filtering algorithm introducing Z-axis height; and if the illumination mode is a dark field illumination mode, determining that the target denoising processing algorithm is a CLAHE algorithm. As an optional implementation manner, the denoising processing of the original image according to the multi-mode scanning information by adopting the target denoising processing algorithm to obtain a processed image includes: Acquiring a template image of the wafer to be tested in the bright field illumination mode; And according to the template image and the Z-axis height when the original image is shot, adopting the guide filtering algorithm for introducing the Z-axis height to perform linear filtering processing on the original image, and obtaining the processed image. As an optional implementation manner, the performing linear filtering processing on the original image by using the guided filtering algorithm for introducing the Z-axis height according to the template image and the Z-axis height when the original image is captured includes: determining a guide image of the original image according to the template image; Determining the defocus height according to the Z-axis height and a preset target focal plane height; In a local window corresponding to the original image, determining a minimized guide filtering cost function corresponding to the local window according to the guide image, the defocus height and the original image by taking the error minimization of the processed image and the original image in the