Low-illuminance image enhancement method and system based on image repair technology

An image enhancement and low-light technology, applied in the field of image processing, can solve the problems of halo at the edge, noise that cannot be removed, and strong noise that is difficult to enhance and restore, so as to achieve good visual effects and solve the effect of missing details

Active Publication Date: 2021-11-26
中科方寸知微(南京)科技有限公司
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In order to improve the visibility of the image and restore the missing details in the image, traditional image enhancement algorithms have achieved initial success in enhancing and restoring low-light images in the past few decades, but due to the limitations of traditional algorithms for complex scenes As well as strong noise and missing details in extremely low-light scenes are difficult to enhance and restore
Despite the rapid development of deep learning technology, low-light image enhancement algorithms based on deep learning have been widely used in this field, but most current low-light enhancement methods still cannot solve the problem of missing details caused by noise.
[0003] In existing technical solutions, for example, although histogram equalization can adjust the brightness of the global image, it cannot remove noise and may reduce contrast
The algorithm based on retinex theory decomposes the image into reflection map and illuminance map, but it is prone to problems such as halo at the edge

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Low-illuminance image enhancement method and system based on image repair technology
  • Low-illuminance image enhancement method and system based on image repair technology
  • Low-illuminance image enhancement method and system based on image repair technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0065] In existing technical solutions, for example, although histogram equalization can adjust the brightness of the global image, it cannot remove noise and may reduce contrast. The algorithm based on the retinex theory decomposes the image into a reflection map and an illumination map, but it is prone to problems such as halos at the edges. In recent years, with the rapid development of deep learning, many algorithms based on deep learning have been proposed one after another. Although they can achieve good results in denoising and brightness adjustment, it is really difficult to recover the lack of details caused by noise.

[0066] For this reason, Embodiment 1 proposes a noise map-guided image repair network for low-illuminance image enhancement, and the technical solution is as follows:

[0067] Prepare the data set of low illumination image, used LOL data set and SYN data set among the present invention;

[0068] Preprocessing the data set, including pixel value normal...

Embodiment 2

[0081] On the basis of Embodiment 1, the applicant further researched and found that most existing low-illuminance image enhancement algorithms cannot remove noise well, and it is difficult to restore the problem of missing details caused by noise. Aiming at the problem of detail loss and difficulty in restoration, we propose a low-light image enhancement algorithm based on image inpainting technology.

[0082] The whole network structure diagram is as follows figure 1 As shown, it consists of a decomposition network and a restoration network in total. The decomposition network consists of three branches, which are responsible for generating reflection maps, illumination maps and noise maps, respectively. The restoration network is composed of Feature Enhancement Group (FEG) and Inpainting Module (InpaintingModule), such as image 3 shown. Each FEG is composed of 4 selection core enhancement modules (SKE), such as figure 2 shown. The decomposition network decomposes the ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The present invention relates to a low-illuminance image enhancement method and system based on image repair technology. The steps of the image enhancement method are as follows: step 1, collect image data, and preprocess the image data; step 2, construct a decomposition network model, and preprocess After the image data is imported into the decomposed network model; Step 3, generate a noise map Mask; Step 4, build a restoration network, and carry out color enhancement and detail restoration to the decomposed image data; Step 5, build a selection kernel enhancement module, Expand the receptive field of the image; step 6, build an image repair module to repair image holes and expand effective information. The invention effectively integrates image repair technology and low-illuminance image restoration, solves the problem of missing details caused by noise, and can repair lost detail information while removing noise to obtain better visual effects.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a low-illuminance image enhancement method and system based on image repair technology. Background technique [0002] Camera Imaging in Dark Light Environment In daily photography, due to dim light, insufficient illumination and insufficient light input of imaging equipment, the generated image usually produces a lot of noise, color degradation, low contrast and underexposure, etc. question. At the same time, this situation also widely exists in other tasks, such as target detection, face recognition, underwater image imaging and video surveillance. In order to improve the visibility of the image and restore the missing details in the image, traditional image enhancement algorithms have achieved initial success in enhancing and restoring low-light images in the past few decades, but due to the limitations of traditional algorithms for complex scenes As well as strong noise and m...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06T5/00G06N3/04G06N3/08
CPCG06T5/002G06T5/007G06N3/08G06T2207/10004G06T2207/20081G06T2207/20084G06N3/048G06N3/045
Inventor 江卓龙冷聪
Owner 中科方寸知微(南京)科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap