Super-resolution method and system, server, user device and method therefor

Patent metadata

Field	Value
Publication number	US-10565684-B2
Application number	US-201615526138-A
Country	US
Kind code	B2
Filing date	Feb 3, 2016
Priority date	Sep 6, 2015
Publication date	Feb 18, 2020
Grant date	Feb 18, 2020

Abstract

A super-resolution method, a super-resolution system, a user equipment and a server. The super-resolution method includes: training an image sample at a server; obtaining a server-side image database; updating a local image database in a user device by using the server-side image database; and displaying an input low-resolution image as a high-resolution image.

First claim

The invention claimed is:

1. A super-resolution method, comprising:
training image samples at a server, including:
receiving new image samples, and making a selection on the new image samples to select image samples among the new image samples;
obtaining a category for each of the selected image samples; and
performing training according to the category for each of the selected image samples to obtain a new parameter or a new set of parameters of the category for each of the selected image samples;

obtaining a server-side image database, including:
updating a template vector in the server-side image database and a low-resolution-high-resolution dictionary block in the server-side image database by using each of the selected image samples and the new parameter or the new set of parameters of the category of each of the selected image samples;

using the server-side image database, updating a local image database in a user device by updating a template vector in the local image database and a low-resolution-high-resolution dictionary block in the local image database based on the updated template vector in the server-side image database and the low-resolution-high-resolution dictionary block in the server-side image database, wherein the template vector in the server-side image database and the template vector in the local image database include the category and the new parameter or the new set of parameters of the category; and
displaying an input low-resolution image as a high-resolution image.

2. The super-resolution method according to claim 1 , wherein the template vector in the server-side image database and the template vector in the local image database have different levels.

3. The super-resolution method according to claim 1 , wherein the image samples are images or a video.

4. A method for performing super-resolution in a user device, comprising:
receiving update data from a server and updating a template vector in a local image database in the user device and a low-resolution-high-resolution dictionary block in the local image database, wherein the template vector includes a category and parameters corresponding to the category, wherein the update data is obtained by the server performing the following steps:
receiving new image samples, and making a selection on the new image samples to select image samples among the new image samples;
obtaining a category for each of the selected image samples;
performing training according to the category for each of the selected image samples to obtain a new parameter or a new set of parameters of the category for each of the selected image samples; and
updating a template vector in a server-side image database and a low-resolution-high-resolutions dictionary block in the server-side side image database by using each of the selected image samples and the new parameter or the new set of parameters of the category of each of the selected image samples, wherein the updated template vector in the server-side image database and the low-resolution-high-resolution dictionary block in the server-side image database are served as the update data; and

based on the local image database, displaying an input low-resolution image as a high-resolution image.

5. The method according to claim 4 , wherein the displaying the input low-resolution image as the high-resolution image comprises:
preprocessing the input low-resolution image;
performing categorization and feature extraction on the preprocessed low-resolution image;
matching the low-resolution image based on the local image database; and
performing an image fusion on the matched low-resolution image to display the high-resolution image.

6. The method according to claim 4 , wherein the template vector in the server-side image database and the template vector in the local image database have different levels, and
wherein updating the template vector in the local image database includes selecting a level of the template vector to perform updating.

7. A method for performing super-resolution on a server, comprising:
training image samples at the server, including:
receiving new image samples, and performing selection on the new image samples to select image samples among the new image samples;
obtaining a category for each of the selected image samples, and
performing training according to the category for each of the selected image samples to obtain a new parameter or a new set of parameters of the category for each of the selected image samples;

obtaining a server-side image database, including:
updating a template vector in the server-side image database and a low-resolution-high-resolution dictionary block in the server-side image database by using each of the selected image samples and the new parameter or the new set of parameters of the category of each of the selected image samples; and

transmitting update data to a user device to update a template vector in a local image database in the user device and a low-resolution-high-resolution dictionary block in the local image database based on the updated template vector in the server-side image database and the low-resolution-high-resolution dictionary block in the server-side image database, enabling the user device to display an input low-resolution image as a high-resolution image based on the server-side image database, wherein the template vector in the server-side image database and the template vector in the local image database include the category and the new parameter or the new set of parameters of the category.

8. The method according to claim 7 , wherein the template vector in the server-side image database and the template vector in the local image database have different levels.

9. The method according to claim 7 , wherein the image samples are images or a video.

10. A super-resolution system comprising a server and a user device, configured to perform the super-resolution method according to claim 1 .

11. A user device, comprising a processor and a memory, wherein the memory stores instructions that, when executed by the processor, cause the user device to perform the method according to claim 4 .

12. A server, comprising a processor and a memory, wherein the memory stores instructions that, when executed by the processor, cause the server to perform the method according to claim 7 .

Assignees

• Boe Technology Group Co Ltd

Inventors

• Zhang Lijie
• He Jianmin
• Michelini Pablo Navarrete
• Zhao Xingxing

Classifications

• G06V10/764

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

• G06T3/4053 Primary

  based on super-resolution, i.e. the output image resolution being higher than the sensor resolution · CPC title

• G06F16/51

  Indexing; Data structures therefor; Storage structures · CPC title

• G06F16/56

  having vectorial format · CPC title

• G06F16/5838

  using colour · CPC title