Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A self-supervised low-light image enhancement and denoising method based on deep learning

An image enhancement and deep learning technology, applied in image enhancement, image analysis, image data processing and other directions, can solve problems such as difficulty in suppressing noise and inability to ensure good contrast, achieve noise suppression, control the uncertainty of enhancement results, suppress The effect of noise information

Active Publication Date: 2021-10-01
HARBIN INST OF TECH
View PDF12 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] In order to solve the problems existing in the existing self-supervised low-illuminance image enhancement methods, such as the inability to ensure that the enhanced result has good contrast and difficulty in suppressing noise, the present invention proposes a self-supervised low-illuminance image enhancement and denoising method based on deep learning

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
  • A self-supervised low-light image enhancement and denoising method based on deep learning
  • A self-supervised low-light image enhancement and denoising method based on deep learning
  • A self-supervised low-light image enhancement and denoising method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0059] Specific implementation mode one: combine figure 1 To describe this embodiment,

[0060] A self-supervised low-light image enhancement and denoising method based on deep learning, comprising the following steps:

[0061] Step 1. Design an image enhancement network based on deep learning, which can decompose a low-light image into a reflection image R and an illumination image I;

[0062] The low-illuminance image S′ to be enhanced and its maximum channel image S′ max And its expected maximum channel image S' expect_max The matrix merged into M*N*5 is used as the input of the self-supervised low-light image enhancement network, and the low-light image is enhanced and denoised using the trained self-supervised low-light image enhancement network, and the output is the reflection image R and the illumination image I. The reflection image R output by the self-supervised low-light image enhancement network is the enhanced image.

[0063] The low-light image S' to be en...

specific Embodiment approach 2

[0099] A self-supervised low-light image enhancement and denoising method based on deep learning described in this embodiment, the specific process of the third step includes the following steps:

[0100] Step 31. Extract the maximum value channel image S of the low-illumination image S max

[0101]

[0102] Among them, S max (i, j) is the maximum channel image S max The i-th row and the j-th column element; max represents the maximum value operation; c is r, g, b, corresponding to the three color channels of red, green and blue in the rgb color space, S c (i,j) is the i-th row and j-th column element of a certain channel of the low-illuminance image S in the rgb color space;

[0103] Step 32, use any contrast enhancement method such as Gamma transform to the maximum value channel image S max Do a contrast enhancement operation to get the desired maximum channel image S expect_max .

[0104] Other steps and parameters are the same as those in the first embodiment.

specific Embodiment approach 3

[0105] Specific implementation mode three: combination figure 2 To describe this embodiment,

[0106] A self-supervised low-light image enhancement and denoising method based on deep learning described in this embodiment, the image enhancement network is specifically as follows:

[0107] The input is input to the first convolutional layer and the second convolutional layer respectively. The first convolutional layer and the second convolutional layer are respectively connected to a 9*9 convolutional layer and connected to an LReLU layer and a 3*3 convolutional layer. A LReLU layer;

[0108] The first convolutional layer is connected to the third convolutional unit, and the third convolutional unit is connected to an LReLU layer after the convolutional layer of 3*3;

[0109] The third convolution unit is connected to the fourth convolution unit, the fourth convolution unit is connected to the fifth convolution unit, the fifth convolution unit is connected to the sixth convol...

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

A self-supervised low-light image enhancement and denoising method based on deep learning belongs to the field of digital image processing. The invention aims to solve the problems that the existing self-supervised low-light image enhancement method based on deep learning is difficult to suppress noise and cannot directly adjust the contrast of the enhanced image. The present invention includes a self-supervised low-light image enhancement network and a regular term for noise suppression. The network can be combined with any existing contrast adjustment method such as Gamma transform to realize network self-supervision training. The noise suppression regular term can be used for The loss function during network training to make the network have noise suppression ability. The invention can preserve color and detail information and significantly suppress noise while enhancing the contrast and brightness of low-illuminance images. The invention can be used for enhancement and denoising of low-illuminance images.

Description

technical field [0001] The invention belongs to the field of digital image processing and relates to methods for enhancing and denoising low-illuminance images. Background technique [0002] Images acquired at night or in dark indoor and other low-light environments often have problems such as low contrast, low brightness, and high noise. In recent years, researchers have proposed a variety of different image enhancement methods, including traditional methods and deep learning-based methods. [0003] Traditional methods include histogram equalization, Gamma transformation, methods based on Retinex theory and improved methods based on these methods, etc. These methods often focus on improving the contrast and brightness of the image, and cannot suppress noise well, and even bring noise Issues such as magnification and color distortion. [0004] Methods based on deep learning can be divided into two types: unsupervised and supervised, and self-supervised methods are unsuperv...

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): G06T5/00G06N3/04
CPCG06T5/002G06T2207/20081G06N3/045
Inventor 张雨王春晖遆晓光李青岩闫诗雨张斌杨国辉崔天祥
Owner HARBIN INST OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com