Determination method, elimination method and apparatus for electron microscope aberration

Patent metadata

Field	Value
Publication number	US-2021286972-A1
Application number	US-202117331393-A
Country	US
Kind code	A1
Filing date	May 26, 2021
Priority date	May 31, 2019
Publication date	Sep 16, 2021
Grant date	—

Abstract

The present disclosure provides a determination method, an elimination method and an apparatus for an electron microscope aberration. The determination method comprises: training a neural network for image recognition using a plurality of electron microscope simulation images to obtain an electron microscope image recognition model; recognizing an electron microscope image of an experimental sample using the electron microscope image recognition model to obtain the electron microscope simulation image corresponding to the electron microscope image of the experimental sample; and obtaining the corresponding set aberration as an imaging aberration of the electron microscope image of the experimental sample according to the electron microscope simulation image corresponding to the electron microscope image of the experimental sample. Through the above solution, an aberration value of an electron microscope can be obtained using a lattice image of the experimental sample, thereby improving an imaging effect of the electron microscope.

First claim

1 . A method for determining an electron microscope aberration, comprising:
training a neural network for image recognition using a plurality of electron microscope simulation images to obtain an electron microscope image recognition model, wherein each of the electron microscope simulation images is corresponding to a respective set aberration; recognizing an electron microscope image of an experimental sample using the electron microscope image recognition model to obtain the electron microscope simulation image corresponding to the electron microscope image of the experimental sample, wherein the electron microscope simulation image and the electron microscope image of the experimental sample are both images of set type; and obtaining the corresponding set aberration as an imaging aberration of the electron microscope image of the experimental sample according to the electron microscope simulation image corresponding to the electron microscope image of the experimental sample.

2 . The method for determining the electron microscope aberration according to claim 1 , wherein training a neural network for image recognition using a plurality of electron microscope simulation images to obtain an electron microscope image recognition model comprises:
performing feature extraction on the electron microscope simulation images to obtain image features; performing cluster analysis on all of the image features to classify all of the electron microscope simulation images; the classified electron microscope simulation images being corresponding to respective categories; and training the neural network for image recognition using the classified electron microscope simulation images based on set network parameters, set network hyper-parameters and a set loss function, to obtain the electron microscope image recognition model.

3 . The method for determining the electron microscope aberration according to claim 2 , wherein before training the neural network for image recognition using the classified electron microscope simulation images based on set network parameters, set network hyper-parameters and a set loss function, to obtain the electron microscope image recognition model, the method further comprises:
performing parameter transfer on network parameters trained based on an ImageNet data set for the images of set type to obtain the set network parameters.

4 . The method for determining the electron microscope aberration according to claim 3 , wherein performing feature extraction on the electron microscope simulation images to obtain image features comprises:
performing the feature extraction on the electron microscope simulation images using the neural network based on the set network parameters, to obtain the image features.

5 . The method for determining the electron microscope aberration according to claim 2 , wherein recognizing an electron microscope image of an experimental sample using the electron microscope image recognition model to obtain the electron microscope simulation image corresponding to the electron microscope image of the experimental sample comprises:
performing category recognition on the electron microscope image of the experimental sample using the electron microscope image recognition model to obtain the category corresponding to the electron microscope image of the experimental sample; and performing image recognition in the category corresponding to the electron microscope image of the experimental sample using the electron microscope image recognition model to obtain the electron microscope simulation image corresponding to the electron microscope image of the experimental sample.

6 . The method for determining the electron microscope aberration according to claim 1 , wherein
the images of set type are high-resolution images; and/or the electron microscope simulation images are transmission electron microscope simulation images; and/or all of the electron microscope simulation images are simulation images corresponding to a same material contained in the experimental sample.

7 . A method for eliminating an electron microscope aberration, comprising:
determining an imaging aberration of an electron microscope image of an experimental sample using the method for determining the electron microscope aberration according to claim 1 ; and eliminating an aberration in the electron microscope image of the experimental sample according to the imaging aberration.

8 . The method for eliminating the electron microscope aberration according to claim 7 , wherein training a neural network for image recognition using a plurality of electron microscope simulation images to obtain an electron microscope image recognition model comprises:
performing feature extraction on the electron microscope simulation images to obtain image features; performing cluster analysis on all of the image features to classify all of the electron microscope simulation images; the classified electron microscope simulation images being corresponding to respective categories; and training the neural network for image recognition using the classified electron microscope simulation images based on set network parameters, set network hyper-parameters and a set loss function, to obtain the electron microscope image recognition model.

9 . The method for eliminating the electron microscope aberration according to claim 8 , wherein before training the neural network for image recognition using the classified electron microscope simulation images based on set network parameters, set network hyper-parameters and a set loss function, to obtain the electron microscope image recognition model, the method further comprises:
performing parameter transfer on network parameters trained based on an ImageNet data set for the images of set type to obtain the set network parameters.

10 . The method for eliminating the electron microscope aberration according to claim 9 , wherein performing feature extraction on the electron microscope simulation images to obtain image features comprises:
performing the feature extraction on the electron microscope simulation images using the neural network based on the set network parameters, to obtain the image features.

11 . The method for eliminating the electron microscope aberration according to claim 8 , wherein recognizing an electron microscope image of an experimental sample using the electron microscope image recognition model to obtain the electron microscope simulation image corresponding to the electron microscope image of the experimental sample comprises:
performing category recognition on the electron microscope image of the experimental sample using the electron microscope image recognition model to obtain the category corresponding to the electron microscope image of the experimental sample; and performing image recognition in the category corresponding to the electron microscope image of the experimental sample using the electron microscope image recognition model to obtain the electron microscope simulation image corresponding to the electron microscope image of the experimental sample.

12 . The method for eliminating the electron microscope aberration according to claim 7 , wherein
the images of set type are high-resolution images; and/or the electron microscope simulation images are transmission electron microscope simulation images; and/or all of the electron microscope simulation images are simulation images corresponding to a same material contained in the experimental sample.

13 . A computer readable storage medium in which a computer program is stored, wherein when being executed by a processor, the

Assignees

• Univ South China Agricult

Inventors

• Lin Fang
• Zhang Qi
• Wang Chen

Classifications

• G01N23/04 Primary

  and forming images of the material · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06V20/698

  Matching; Classification · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title