Search

CN-121999345-A - Image quality evaluation method and SEM focusing method for SEM image

CN121999345ACN 121999345 ACN121999345 ACN 121999345ACN-121999345-A

Abstract

The embodiment of the invention discloses an image quality evaluation method and an SEM focusing method of an SEM image, wherein the image quality evaluation method comprises the steps of filtering low-frequency noise in a first image to obtain a second image, wherein the first image is the SEM image; and obtaining the difference degree of the second image and the third image through a pretrained twin neural network, and determining the image quality evaluation score of the first image based on the difference degree. The embodiment of the invention can improve the evaluation accuracy when evaluating the image quality of the SEM image with low-frequency noise and high-frequency noise.

Inventors

  • LI YIQING

Assignees

  • 上海精测半导体技术有限公司

Dates

Publication Date
20260508
Application Date
20241106

Claims (10)

  1. 1. An image quality evaluation method of an SEM image, comprising: Filtering low-frequency noise in a first image to obtain a second image, wherein the first image is an SEM image; Performing fuzzy degradation processing on the second image to obtain a third image, and And acquiring the difference degree of the second image and the third image through a pretrained twin neural network, and determining the image quality evaluation score of the first image based on the difference degree.
  2. 2. The method for evaluating image quality of SEM image according to claim 1, wherein said filtering low frequency noise in the first image to obtain the second image includes: performing Fourier transform on the first image to obtain a first spectrogram; Filtering the low-frequency noise signal in the first spectrogram by adopting a high-pass filtering method to obtain a second spectrogram, and And carrying out inverse Fourier transform on the second spectrogram to obtain the second image.
  3. 3. The method for evaluating image quality of SEM image according to claim 2, wherein said filtering the low-frequency noise signal in the first spectrogram by high-pass filtering to obtain a second spectrogram comprises: carrying out frequency spectrum centering treatment on the first spectrogram to obtain a centering spectrogram, wherein the central area of the centering spectrogram is a low-frequency noise signal corresponding to low-frequency noise in a first image; Constructing a mask image of the same size as the centered spectrogram and such that the mask image includes a 0-valued area and a 1-valued area, wherein the value of each image point within the 0-valued area is 0, the 0-valued area is located at the center of the mask image, the value of each image point within the 1-valued area is 1, the 1-valued area is located at the periphery of the 0-valued area in the mask image, and And filtering the low-frequency noise signal in the centralized spectrogram by using the mask image to obtain the second spectrogram.
  4. 4. The image quality evaluation method of an SEM image according to claim 3, wherein the mask image further includes a transition region, wherein the transition region is located between the 0-value region and the 1-value region, and wherein each of the image points in the transition region has a value greater than 0 and less than 1.
  5. 5. The image quality evaluation method of an SEM image according to claim 4, wherein said constructing a mask image of the same size as said centered spectrogram includes: Acquiring a first distance between an inner boundary of the transition region and a center point of a mask image and a second distance between an outer boundary of the transition region and the center point of the mask image, and determining a position of the transition region in the mask image based on the first distance and the second distance, and The mask image is constructed based on the location of the transition region in the mask image.
  6. 6. The method for evaluating image quality of an SEM image according to claim 4 or 5, The values of the individual image points in the transition region increase gradually in a linear manner with increasing distance of the respective image point from the center point of the mask image.
  7. 7. The image quality evaluation method of SEM image according to claim 6, wherein the calculation formula of the values of the respective image points of the mask image is: Where M represents the value of each image point of the mask image, d center represents the distance from the image point to the center point of the mask image, th 1 represents the first distance from the inner boundary of the transition region to the center point of the mask image, and th 2 represents the second distance from the outer boundary of the transition region to the center point of the mask image.
  8. 8. The image quality evaluation method of an SEM image according to claim 1, wherein said subjecting the second image to the blur degradation process to obtain a third image includes: performing fuzzy degradation processing on the second image based on the mean value fuzzy check to obtain a third image; the determining an image quality evaluation score of the first image based on the degree of difference includes: And directly determining the difference degree as the image quality evaluation score, or multiplying the difference degree by a preset proportionality coefficient to obtain the image quality evaluation score.
  9. 9. The image quality evaluation method of an SEM image according to claim 1, wherein the twin neural network includes a first neural network, a second neural network, and a contrast loss layer; The obtaining the difference degree of the second image and the third image through the pretrained twin neural network comprises the following steps: inputting the second image into the first neural network, and extracting the characteristics of the second image through the first neural network to obtain a first characteristic vector; inputting the third image into the second neural network, extracting the features of the third image through the second neural network to obtain a second feature vector, and Determining, by the contrast loss layer, a degree of difference of the second image and the third image based on the first feature vector and the second feature vector.
  10. 10. A SEM focusing method, comprising: acquiring a plurality of first images, determining image quality evaluation scores of the plurality of first images by the image quality evaluation method of the SEM image according to any one of claims 1 to 9; acquiring the first image corresponding to the highest score of all the image quality evaluation scores as a target image, and Focusing the corresponding SEM based on the imaging parameters corresponding to the target image.

Description

Image quality evaluation method and SEM focusing method for SEM image Technical Field The invention relates to the technical field of image processing, in particular to an image quality evaluation method and an SEM focusing method of an SEM image. Background Scanning electron microscopy (Scanning Electron Microscope, SEM) uses an electron beam to scan a sample, such as a wafer, to obtain a high resolution SEM image that can clearly show micro-nano scale structures, which can be used for analysis of surface topography and material properties. In order to obtain a highly accurate and reliable image, an image quality evaluation of SEM images is required. Specifically, by evaluating the image quality of the SEM image, the imaging parameters corresponding to the optimal image, i.e., the target image, may be obtained, or the image quality may be monitored in real time, so as to facilitate automatic detection of possible defects or anomalies in the image, or low-quality images or irrelevant images may be filtered, so that invalid data storage and transmission are reduced. There is often both low frequency noise and high frequency noise in SEM images. For example, a sample having poor conductivity, such as an insulating sample, is liable to cause charge accumulation during electron beam scanning, resulting in low-frequency noise of uneven brightness in an SEM image, and the SEM image tends to be affected by instability of an electron source to introduce high-frequency electron beam noise. However, the image quality evaluation method in the prior art is poor in evaluation accuracy when evaluating SEM images having both low-frequency noise and high-frequency noise. Disclosure of Invention The embodiment of the invention provides an image quality evaluation method and an SEM focusing method for an SEM image, which are used for improving the evaluation accuracy when the image quality of the SEM image with low-frequency noise and high-frequency noise is evaluated. In a first aspect, an embodiment of the present invention provides an image quality evaluation method for an SEM image, including: Filtering low-frequency noise in a first image to obtain a second image, wherein the first image is an SEM image; Performing fuzzy degradation processing on the second image to obtain a third image, and And acquiring the difference degree of the second image and the third image through a pretrained twin neural network, and determining the image quality evaluation score of the first image based on the difference degree. In a second aspect, an embodiment of the present invention provides an SEM focusing method, including: Acquiring a plurality of first images, and determining image quality evaluation scores of the plurality of first images by the image quality evaluation method of the SEM image; acquiring the first image corresponding to the highest score of all the image quality evaluation scores as a target image, and Focusing the corresponding SEM based on the imaging parameters corresponding to the target image. According to the image quality evaluation method and the SEM focusing method for the SEM image, low-frequency noise in the SEM image (namely the first image) is filtered, fuzzy degradation processing is carried out on the SEM image after noise reduction (namely the second image), then the difference degree of the SEM image after noise reduction and fuzzy degradation (namely the third image) is obtained by utilizing the pretrained twin neural network, and the image quality evaluation score of the SEM image is further determined based on the difference degree, so that the evaluation accuracy in image quality evaluation of the SEM image with low-frequency noise and high-frequency noise can be improved, the focusing of the SEM is realized based on the accurate image quality evaluation score, and the accuracy of the focusing method is improved. Drawings In order to more clearly illustrate the technical solutions of the present invention, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present invention and should not be considered as limiting the scope, and that other related drawings can be obtained according to these drawings without inventive effort for a person skilled in the art. Fig. 1 is a schematic flow chart of an image quality evaluation method of SEM images according to an embodiment of the present invention; Fig. 2 is a schematic flow chart of a process of filtering low-frequency noise in a first image in an image quality evaluation method of SEM images according to an embodiment of the present invention; fig. 3 is a schematic flow chart of a process of filtering a low-frequency noise signal in a first spectrogram in an image quality evaluation method of an SEM image according to an embodiment of the present invention; Fig. 4 is a schematic flow chart of a process of acquiring a difference degree