Image generation system

Patent metadata

Field	Value
Publication number	US-11663713-B2
Application number	US-202117155680-A
Country	US
Kind code	B2
Filing date	Jan 22, 2021
Priority date	Feb 5, 2020
Publication date	May 30, 2023
Grant date	May 30, 2023

Abstract

Provided is a system that constructs a learning database for a learning model in a short period of time. The system generates an image in which a structure image is similar to an actual image by image processing of the structure image. One or more processors acquire a first structure image and a second structure image different from the first structure image. The one or more processors create a plurality of intermediate structure images indicating an intermediate structure between the first structure image and the second structure image. The one or more processors generate an image by making each of the plurality of intermediate structure images to be similar to an actual image by image processing of each of the plurality of intermediate structure images.

First claim

What is claimed is:

1. A system that generates an image in which a structure image is similar to an actual image by image processing of the structure image, the system comprising:
one or more storage apparatuses; and
one or more processors that operate in accordance with a program stored in the one or more storage apparatuses, wherein
the one or more processors
acquire a first structure image and a second structure image different from the first structure image,
create a plurality of intermediate structure images indicating an intermediate structure between the first structure image and the second structure image, and
generate an image by making each of the plurality of intermediate structure images to be similar to an actual image by the image processing of each of the plurality of intermediate structure images.

2. The system according to claim 1 , wherein
the one or more processors generate an image in which each of the plurality of intermediate structure images is similar to the actual image based on a reference image.

3. The system according to claim 1 , wherein
one of the first structure image and the second structure image is a structure image corresponding to a goal shape.

4. The system according to claim 1 , wherein
one of the first structure image and the second structure image is a structure image corresponding to a shape closest to a goal shape in an acquired actual image.

5. The system according to claim 1 , wherein
the first structure image is a structure image corresponding to a shape closest to a goal shape in an acquired actual image, and
the second structure image is a structure image corresponding to the goal shape.

6. The system according to claim 1 , wherein
the one or more processors generate the plurality of intermediate structure images by changing a structure image from the first structure image to the second structure image according to a specified correspondence.

7. The system according to claim 1 , wherein
the one or more processors
construct a database including the plurality of intermediate structure images and an image in which each of the plurality of intermediate structure images is similar to the actual image,
construct a structure acquisition model that outputs a structure image from an input actual image by using the database, and
generate a new structure image from a new actual image using the structure acquisition model.

8. The system according to claim 7 , wherein
the one or more processors measure a predetermined feature dimension in the new structure image using a dimension extraction model that extracts the feature dimension.

9. The system according to claim 8 , wherein
the one or more processors
construct a learning model that outputs a structure image corresponding to an input apparatus condition using a database including the measured feature dimension, the new structure image, the new actual image, and an apparatus condition for processing a sample from which the new actual image is acquired, and
estimate an apparatus condition for implementing a goal shape using the learning model.

10. The system according to claim 2 , wherein
one of the first structure image and the second structure image is a structure image corresponding to a goal shape.

11. The system according to claim 2 , wherein
one of the first structure image and the second structure image is a structure image corresponding to a shape closest to a goal shape in an acquired actual image.

12. The system according to claim 2 , wherein
the first structure image is a structure image corresponding to a shape closest to a goal shape in an acquired actual image, and
the second structure image is a structure image corresponding to the goal shape.

13. The system according to claim 2 , wherein
the one or more processors generate the plurality of intermediate structure images by changing a structure image from the first structure image to the second structure image according to a specified correspondence.

14. The system according to claim 2 , wherein
the one or more processors
construct a database including the plurality of intermediate structure images and an image in which each of the plurality of intermediate structure images is similar to the actual image,
construct a structure acquisition model that outputs a structure image from an input actual image by using the database, and
generate a new structure image from a new actual image using the structure acquisition model.

15. The system according to claim 14 , wherein
the one or more processors measure a predetermined feature dimension in the new structure image using a dimension extraction model that extracts the feature dimension.

16. The system according to claim 15 , wherein
the one or more processors
construct a learning model that outputs a structure image corresponding to an input apparatus condition using a database including the measured feature dimension, the new structure image, the new actual image, and an apparatus condition for processing a sample from which the new actual image is acquired, and
estimate an apparatus condition for implementing a goal shape using the learning model.

17. A method for a system to generate an image in which a structure image is similar to an actual image by image processing of the structure image, the method comprising:
acquiring, by the system, a first structure image and a second structure image different from the first structure image,
creating, by the system, a plurality of intermediate structure images indicating an intermediate structure between the first structure image and the second structure image, and
generating, by the system, an image by making each of the plurality of intermediate structure images to be similar to an actual image by image processing of each of the plurality of intermediate structure images.

Assignees

• Hitachi Ltd

Inventors

• Ohmori Takeshi
• Okuyama Yutaka

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0475

  Generative networks · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/045

  Combinations of networks · CPC title