CN-120672664-B - Wafer image preprocessing method, device, equipment and storage medium

CN120672664BCN 120672664 BCN120672664 BCN 120672664BCN-120672664-B

Abstract

The invention discloses a wafer image preprocessing method, device, equipment and storage medium, wherein the method comprises the steps of obtaining an initial to-be-detected image of a wafer to be detected, carrying out noise distribution estimation on the initial to-be-detected image, locating a highlight defect area in the initial to-be-detected image, carrying out smoothing treatment on the highlight defect area to obtain a smooth image and a difference image of the highlight defect area and a corresponding area on the smooth image, carrying out frequency domain filtering operation on the smooth image to obtain a filtered image, and overlapping the difference image to the corresponding area on the filtered image to obtain a final to-be-detected image. According to the invention, the highlight defect area in the wafer diffraction image is positioned and separated, the other part of the image is subjected to frequency domain filtering operation, and finally, the difference image corresponding to the highlight defect area is overlapped to the filtered image, so that image reconstruction is completed, the morphological characteristics of the highlight defect are reserved, and the detection and classification of the follow-up defect are not influenced.

Inventors

ZHANG LONG
DU SHUANG
ZHU YANMING
CHEN LU

Assignees

广州中科飞测科技有限公司

Dates

Publication Date: 20260508
Application Date: 20250521

Claims (9)

1. A method for preprocessing a wafer image, comprising: acquiring an initial image to be inspected of a wafer to be inspected; performing noise distribution estimation on the initial to-be-detected image to calculate a pixel threshold value of each pixel point based on a reference pixel value in a preset neighborhood range of the pixel point, and taking an area formed by the pixel points with the pixel values exceeding the pixel threshold value as a highlight defect area in the initial to-be-detected image; Carrying out smoothing treatment on the highlight defect region to obtain a smooth image and a difference image of the highlight defect region and a corresponding region on the smooth image; performing frequency domain filtering operation on the smooth image to obtain a filtered image; and overlapping the difference image to a corresponding area on the filtering image to obtain a final image to be detected.
2. The method for preprocessing a wafer image according to claim 1, wherein calculating a pixel threshold value of each pixel based on a reference pixel value within a preset neighborhood range of the pixel comprises: extracting reference pixel values of all reference pixel points in a preset neighborhood range of each pixel point; calculating a first characteristic value and a second characteristic value of each pixel point by using the reference pixel value; A pixel threshold value for each pixel is calculated based on the first feature value and the second feature value.
3. The method according to claim 2, wherein the first characteristic value includes a median value or a mean value of all reference pixel values in a preset neighborhood range, and the second characteristic value includes a first characteristic value of all reference pixel values in the preset neighborhood range; The pixel threshold calculation process is expressed as: ; Wherein, the Represent the first A pixel threshold value of each pixel point, Represent the first The median or mean value corresponding to each pixel point, Represent the first First variance values corresponding to the pixel points, The table is a preset noise estimation coefficient.
4. The method of claim 2, wherein the first feature value comprises a median or mean value of all reference pixel values in a preset neighborhood range, and the second feature value comprises a second variance value of all reference pixel values in the preset neighborhood range, the second variance value being expressed as: ; Wherein, the Represent the first A second variance value corresponding to each pixel point, Represent the first The median or mean value corresponding to each pixel point, Represent the first Minimum pixel values in a preset neighborhood range corresponding to the pixel points, Representing pre-calibrated parameters; The pixel threshold calculation process is expressed as: [ ]; Wherein, the Represent the first A pixel threshold value of each pixel point, The table is a preset noise estimation coefficient.
5. The method according to claim 4, wherein the preset neighborhood range is represented by all pixels within a window with a size of M x N, which is constructed by taking the pixels as a center.
6. The method for preprocessing a wafer image according to claim 5, wherein said smoothing the highlight defect region to obtain a smoothed image and a difference image between the highlight defect region and a corresponding region on the smoothed image, comprises: Obtaining target pixel values of all target pixel points in the highlight defect area; Calculating a difference value between a target pixel value corresponding to each target pixel point and a corresponding first characteristic value, and constructing a difference image based on the difference value; and replacing the pixel value of each target pixel point in the highlight defect area with a corresponding first characteristic value to obtain the smooth image.
7. A wafer image preprocessing apparatus, comprising: the acquisition module is used for acquiring an initial image to be detected of the wafer to be detected; The positioning module is used for carrying out noise distribution estimation on the initial image to be detected, calculating a pixel threshold value of each pixel point based on a reference pixel value in a preset neighborhood range of the pixel point, and taking an area formed by the pixel points with the pixel values exceeding the pixel threshold value as a highlight defect area in the initial image to be detected; The smoothing processing module is used for carrying out smoothing processing on the highlight defect area to obtain a smooth image and a difference image of the highlight defect area and a corresponding area on the smooth image; The filtering module is used for carrying out frequency domain filtering operation on the smooth image to obtain a filtered image; And the reconstruction module is used for overlapping the difference image to a corresponding area on the filtering image to obtain a final image to be detected.
8. A computer device comprising a processor, a memory coupled to the processor, the memory having stored therein program instructions that, when executed by the processor, cause the processor to perform the steps of the wafer image preprocessing method of any one of claims 1-6.
9. A storage medium storing program instructions capable of implementing the wafer image preprocessing method according to any one of claims 1 to 6.

Description

Wafer image preprocessing method, device, equipment and storage medium Technical Field The present application relates to the field of wafer inspection technologies, and in particular, to a wafer image preprocessing method, a device, equipment, and a storage medium. Background The wafer generates reflected or scattered light under the irradiation of the laser light spots, the reflected or scattered light is collected by the imaging light path and then irradiates on the photosensitive light source of the camera to form an image with wafer background, structure and defect information, and the yield of the wafer can be judged by detecting the defects of the image. The imaging image generally contains noise introduced by laser, wafer process and camera besides background, structure and defect information, the noise can interfere with defect detection, the periodic noise filtering is usually completed in a frequency domain, but the image which is reversely transformed after the frequency domain filtering is usually generated in a highlight area to damage structural characteristics of defects, and the follow-up defect classification flow is influenced. At present, for the problem of image distortion in a highlight region, a common practice is to superimpose a gaussian filter or other smooth filters on the filter, and the method can reduce the distortion effect of the image after inverse transformation to a certain extent, but the smoothing of the filter can influence the effect of frequency domain filtering to a certain extent, and the distortion phenomenon of the highlight defect is weakened, so that the effect is insufficient, and the defect detection accuracy of the wafer is still greatly influenced. Disclosure of Invention In view of the above, the present application provides a method, apparatus, device and storage medium for preprocessing a wafer image, so as to solve the problem of distortion of a highlight defect image in the existing preprocessing of a wafer to-be-inspected image. The wafer image preprocessing method comprises the steps of obtaining an initial to-be-detected image of a wafer to be detected, carrying out noise distribution estimation on the initial to-be-detected image, locating a highlight defect area in the initial to-be-detected image, carrying out smoothing treatment on the highlight defect area to obtain a smooth image and a difference image of the highlight defect area and a corresponding area on the smooth image, carrying out frequency domain filtering operation on the smooth image to obtain a filtered image, and overlapping the difference image to the corresponding area on the filtered image to obtain a final to-be-detected image. The method comprises the steps of extracting reference pixel values of all reference pixel points in a preset neighborhood range of each pixel point, calculating a first characteristic value and a second characteristic value of each pixel point by using the reference pixel values, calculating a pixel threshold value of each pixel on the basis of the first characteristic value and the second characteristic value, and taking a region formed by pixels with the pixel values exceeding the pixel threshold value as a highlight defect region. As a further improvement of the present application, the first characteristic value includes a median value or a mean value of all reference pixel values in a preset neighborhood range, the second characteristic value includes a first characteristic value of all reference pixel values in the preset neighborhood range, and the calculation process of the pixel threshold value is expressed as: Ti＝medaini+coeff1*stdi; Wherein T i represents a pixel threshold of the ith pixel, medain i represents a median or mean value corresponding to the ith pixel, std i represents a first variance value corresponding to the ith pixel, and coeff 1 represents a noise estimation coefficient set in advance. As a further refinement of the application, the first characteristic value comprises a median or mean value of all reference pixel values within a preset neighborhood range, the second characteristic value comprises a second variance value of all reference pixel values within the preset neighborhood range, the second variance value being expressed as: stdj＝coeff2*(medainj-minj); Wherein std j represents a second variance value corresponding to the jth pixel, medain j represents a median or average value corresponding to the jth pixel, min j represents a minimum pixel value in a preset neighborhood range corresponding to the jth pixel, and coeff 2 represents a pre-calibration parameter; The pixel threshold calculation process is expressed as: Tj＝medainj+coeff1*[coeff2*(medainj-minj)]; wherein, T j represents the pixel threshold of the j-th pixel, and coeff 1 represents the noise estimation coefficient set in advance. As a further improvement of the present application, the preset neighborhood range is expressed as all pixel points within a wind