System, method, and storage medium for converting low-resolution image data into high-resolution image data using neural network

Patent metadata

Field	Value
Publication number	US-11272074-B2
Application number	US-202016793860-A
Country	US
Kind code	B2
Filing date	Feb 18, 2020
Priority date	Feb 28, 2019
Publication date	Mar 8, 2022
Grant date	Mar 8, 2022

Abstract

A system previously generates a plurality of neural networks by performing learning based on an image data pair generated based on identical original data and differing in resolution. Then, the system determines one neural network out of the plurality of neural networks based on a resolution of input image data and a resolution of an image to be output, acquires post-conversion image data based on the determined neural network and the input image data, and performs outputting that is based on the post-conversion image data.

First claim

What is claimed is:

1. A system comprising:
at least one memory that stores a program; and
at least one processor that executes the program to perform:
acquiring first information about a first resolution of an input image;
acquiring second information about a second resolution of an image that is to be output, wherein the second resolution is higher than the first resolution;
determining, from among a plurality of neural networks that convert a low-resolution image into a high-resolution image, one neural network based on the first information and the second information, wherein the determined one neural network is a neural network suitable for converting an image having the first resolution into an image having the second resolution,
wherein the plurality of neural networks are obtained by performing learning based on an image data pair which includes first image data and second image data which is higher in resolution than the first image data,
wherein the second image data is obtained by rendering page-description language (PDL) data included in a print job, and
wherein the first image data is obtained at least by imparting noises to the obtained second image data and converting the noises-imparted second image data into an image of a low-resolution;
acquiring post-conversion image data having the second resolution into which the input image having the first resolution is converted using the determined one neural network; and
outputting the acquired post-conversion image data.

2. The system according to claim 1 , wherein the first image data is binary data and the second image data is multivalued data.

3. The system according to claim 1 , further comprising a printer,
wherein the outputting includes causing the printer to form the acquired post-conversion image data on a sheet, and
wherein the second information is information about the second resolution at which image formation is performed by the printer.

4. The system according to claim 1 , further comprising a user interface configured to receive an operation performed by a user,
wherein the second information is information about the second resolution designated by the user via the user interface.

5. The system according to claim 1 , wherein the noise imparting processing includes processing for imparting random noise to the obtained second image data.

6. The system according to claim 1 , wherein the first image data is obtained by imparting noises to the obtained second image data, converting the noises-imparted second image data into an image of a low-resolution and further applying binarization processing to the converted image of the low-resolution.

7. The system according to claim 1 , wherein the noise imparting processing includes processing for imparting respective independent random noises to a surrounding region of a foreground region of the obtained second image data and an inside region of the foreground region of obtained the second image data.

8. A method comprising:
acquiring first information about a first resolution of an input image;
acquiring second information about a second resolution of an image that is to be output, wherein the second resolution is higher than the first resolution;
determining, from among a plurality of neural networks that convert a low-resolution image into a high-resolution image, one neural network based on the first information and the second information, wherein the determined one neural network is a neural network suitable for converting an image having the first resolution into an image having the second resolution,
wherein the plurality of neural networks are obtained by performing learning based on an image data pair which includes first image data and second image data which is higher in resolution than the first image data, wherein the second image data is obtained by rendering page-description language (PDL) data included in a print job, and
wherein the first image data is obtained at least by imparting noises to the obtained second image data and converting the noises-imparted second image data into an image of a low-resolution;
acquiring post-conversion image data having the second resolution into which the input image having the first resolution is converted using the determined one neural network; and
outputting the acquired post-conversion image data.

9. The method according to claim 8 ,
wherein the outputting includes causing a printer to form the acquired post-conversion image data on a sheet, and
wherein the second information is information about the second resolution at which image formation is performed by the printer.

10. The method according to claim 8 , further comprising receiving an operation performed by a user by a user interface,
wherein the second information is information about the second resolution designated by the user via the user interface.

11. A non-transitory computer-readable storage medium storing computer-executable instructions that, when executed by a computer, cause the computer to perform a method comprising:
acquiring first information about a first resolution of an input image;
acquiring second information about a second resolution of an image that is to be output, wherein the second resolution is higher than the first resolution;
determining, from among a plurality of neural networks that convert a low-resolution image into a high-resolution image, one neural network based on the first information and the second information, wherein the determined one neural network is a neural network suitable for converting an image having the first resolution into an image having the second resolution,
wherein the plurality of neural networks is obtained by performing learning based on an image data pair which includes first image data and second image data which is higher in resolution than the first image data,
wherein the second image data is obtained by rendering page-description language (PDL) data included in a print job, and
wherein the first image data is obtained at least by imparting noises to the obtained second image data and converting the noises-imparted second image data into an image of a low-resolution;
acquiring post-conversion image data having the second resolution into which the input image having the first resolution is converted using the determined one neural network; and
outputting the acquired post-conversion image data.

12. The non-transitory computer-readable storage medium according to claim 11 ,
wherein the outputting includes causing a printer to form the acquired post-conversion image data on a sheet, and
wherein the second information is information about the second resolution at which image formation is performed by the printer.

13. The non-transitory computer-readable storage medium according to claim 11 , further comprising receiving an operation performed by a user by a user interface,
wherein the second information is information about the second resolution designated by the user via the user interface.

14. A non-transitory computer-readable storage medium storing computer-executable instructions that, when executed by a computer, cause the computer to perform a method comprising:
obtaining second image data by rendering page-description language (PDL) data included in a print job;
obtaining first image data at least by imparting noises to the obtained second image data and converting the noises-imparted second image data into an image of a low-resolution, wherein a resolution of the first image is lower than a resolution of the second image; and
generating a neural network for converting a low-resolution image into a

Assignees

• Canon Kk

Inventors

• Achiwa Ken
• Kosaka Ryo

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• H04N1/409

  Edge or detail enhancement; Noise or error suppression · CPC title

• H04N1/4092

  Edge or detail enhancement · CPC title