Image enhancement method and apparatus

Patent metadata

Field	Value
Publication number	US-12536628-B2
Application number	US-202217687374-A
Country	US
Kind code	B2
Filing date	Mar 4, 2022
Priority date	Sep 4, 2019
Publication date	Jan 27, 2026
Grant date	Jan 27, 2026

Abstract

This application relates to an image enhancement technology in the field of computer vision in the field of artificial intelligence, and provides an image enhancement method and apparatus. This application relates to the field of artificial intelligence, and specifically, to the field of computer vision. The method includes: adjusting a pixel value of a to-be-processed image, to obtain K images, where pixel values of the K images are different, and K is a positive integer greater than 1; extracting local features of the K images; extracting a global feature of the to-be-processed image; and performing image enhancement processing on the to-be-processed image based on the global feature and the local features, to obtain an image-enhanced output image. This method helps to improve the effect of image quality enhancement processing.

First claim

What is claimed is:

1 . An image enhancement method, comprising:
obtaining K images based on a to-be-processed image, wherein K is a positive integer greater than 1, wherein the to-be-processed image comprises a plurality of pixels, wherein each of the K images is obtained by increasing all pixel values or decreasing all pixel values of the to-be-processed image, and wherein corresponding pixels of the K images have different pixel values; performing semantic segmentation on the to-be-processed image, to obtain a semantic segmentation graph; performing feature extraction on the semantic segmentation graph to obtain a semantic condition, wherein the semantic condition comprises semantic information of the to-be-processed image; extracting a set of local features from each image of the K images, wherein, for each image of the K images, at least one local feature of the set of local features is determined based on a residual estimated value, wherein the residual estimated value is determined by a residual block that extracts a corresponding local feature and is conditioned on the semantic condition; extracting a global feature of the to-be-processed image, wherein the global feature is associated with the to-be-processed image; and performing image enhancement processing on the to-be-processed image based on the global feature and the sets of local features of the K images, to obtain an image-enhanced output image.

2 . The method according to claim 1 , wherein the obtaining the K images comprises:
increasing all pixel values or decreasing all pixel values of the to-be-processed image through one or more non-linear transformations.

3 . The method according to claim 1 , wherein at least one local feature of the set of local features is extracted from each image of the K images based on a first semantic feature and a second semantic feature, wherein the first semantic feature and the second semantic feature are determined based on the semantic condition.

4 . The method according to claim 3 , wherein the residual estimated value is determined based on the first semantic feature, the second semantic feature, and an image feature of the to-be-processed image.

5 . The method according to claim 1 , further comprising:
performing feature fusion on the sets of local features of the K images, to obtain a local fusion feature; and wherein the performing image enhancement processing on the to-be-processed image to obtain the image-enhanced output image comprises:
performing image enhancement processing on the to-be-processed image based on the global feature and the local fusion feature, to obtain the image-enhanced output image.

6 . The method according to claim 1 , wherein the performing image enhancement processing on the to-be-processed image to obtain the image-enhanced output image comprises:
performing feature fusion on the global feature and the sets of local features of the K images, to obtain a fusion feature; and performing image enhancement processing on the to-be-processed image based on the fusion feature, to obtain the image-enhanced output image.

7 . The method according to claim 6 , wherein the performing feature fusion on the global feature and the sets of local features of the K images, to obtain the fusion feature comprises:
performing feature extraction on the global feature, to obtain a first global feature and a second global feature; performing addition on the first global feature and the sets of local features, to obtain a candidate fusion feature; and performing concatenation and convolution on the candidate fusion feature and the second global feature, to obtain the fusion feature.

8 . The method according to claim 6 , wherein the performing feature fusion on the global feature and the sets of local features of the K images, to obtain the fusion feature comprises:
performing feature extraction on the global feature, to obtain a first global feature and a second global feature; performing concatenation and convolution on the first global feature and the sets of local features, to obtain a candidate fusion feature; and performing addition on the candidate fusion feature and the second global feature, to obtain the fusion feature.

9 . The method according to claim 1 ,
wherein the to-be-processed image is an image obtained after an original image of the to-be-processed image is downsampled, wherein the performing image enhancement processing on the to-be-processed image to obtain the image-enhanced output image comprises:
performing image enhancement processing on the original image of the to-be-processed image based on the global feature and the sets of local features of the K images, to obtain the image-enhanced output image.

10 . An image enhancement apparatus, comprising:
a processor; and a memory, wherein the memory is configured to store program instructions, and the processor is configured to invoke the program instructions to perform a method comprising:
obtaining K images based on a to-be-processed image, wherein K is a positive integer greater than 1, wherein the to-be-processed image comprises a plurality of pixels, wherein each of the K images is obtained by increasing all pixel values or decreasing all pixel values of the to-be-processed image, and wherein corresponding pixels of the K images have different pixel values;
performing semantic segmentation on the to-be-processed image, to obtain a semantic segmentation graph;
performing feature extraction on the semantic segmentation graph to obtain a semantic condition, wherein the semantic condition comprises semantic information of the to-be-processed image;
extracting a set of local features from each image of the K images, wherein, for each image of the K images, at least one local feature of the set of local features is determined based on a residual estimated value, wherein the residual estimated value is determined by a residual block that extracts a corresponding local feature and is conditioned on the semantic condition;
extracting a global feature of the to-be-processed image, wherein the global feature is associated with the to-be-processed image; and
performing image enhancement processing on the to-be-processed image based on the global feature and the sets of local features of the K images, to obtain an image-enhanced output image.

11 . The apparatus according to claim 10 , wherein the obtaining the K images comprises:
increasing all pixel values or decreasing all pixel values of the to-be-processed image through one or more non-linear transformations.

12 . The apparatus according to claim 10 , wherein at least one local feature of the set of local features is extracted from each image of the K images based on a first semantic feature and a second semantic feature, wherein the first semantic feature and the second semantic feature are determined based on the semantic condition.

13 . The apparatus according to claim 12 , wherein the residual estimated value is determined based on the first semantic feature, the second semantic feature, and an image feature of the to-be-processed image.

14 . The apparatus according to claim 10 , wherein the method further comprises:
performing feature fusion on the sets of local features of the K images, to obtain a local fusion feature; and wherein the performing image enhancement processing on the to-be-processed image to obtain the image-enhanced output image comprises: performing image enhancement processing on the to-be-processed image based on the global feature and the local fusion feature, to obtain the image-enhanced output image.

Assignees

• Huawei Tech Co Ltd

Inventors

• Wang Tao
• Wang Xian
• Song Fenglong

Classifications

• G06T2207/20221

  Image fusion; Image merging · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/20072

  Graph-based image processing · CPC title

• G06V10/82

  using neural networks · CPC title