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

Low-illumination color image enhancement method based on Retinex and convolutional neural network

A convolutional neural network and color image technology, applied in the field of low-light color image enhancement, can solve problems such as difficult to meet actual needs, distortion, noise color, etc.

Active Publication Date: 2019-09-13
TIANJIN UNIV
View PDF7 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these methods cannot effectively deal with the serious noise and color distortion in the enhanced image.
[0007]In summary, most of the existing low-light image enhancement algorithms only have a certain effect on some low-light images without noise, but for some particularly dark areas Enhanced results often have serious noise and color distortion, which is difficult to meet actual needs

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-illumination color image enhancement method based on Retinex and convolutional neural network
  • Low-illumination color image enhancement method based on Retinex and convolutional neural network
  • Low-illumination color image enhancement method based on Retinex and convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0026] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but the following embodiments do not limit the present invention in any way.

[0027] The invention proposes a low-light color image enhancement method based on Retinex and convolutional neural network. The design idea is to design three different convolutional neural networks, which are decomposition network, reflection map restoration network and illumination map adjustment network. The basic steps are: first input the low-light color image to the decomposition network designed by the present invention, and the decomposition network outputs a three-channel reflection map and a single-channel illumination map; then input the reflection map and illumination map to the reflection map restoration network , perform denoising and color restoration processing, and obtain the restored reflection map; then input the light map and light adjustment paramet...

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 invention discloses a low-illumination color image enhancement method based on Retinex and a convolutional neural network. The low-illumination color image enhancement method comprises the steps:inputting a low-illumination color image into a decomposition network, and outputting a three-channel reflection image and a single-channel illumination image; inputting the reflection image and the illumination image into a reflection image recovery network, and carrying out denoising and color recovery processing to obtain a recovered reflection image; inputting the illumination image and the illumination adjustment parameters into an illumination image adjustment network, and outputting the adjusted illumination image; and finally, carrying out dot multiplication operation on the recoveredreflection image and the adjusted illumination image to obtain an enhanced image. Based on the Retinex theory, the low-illumination color image enhancement method utilizes the convolutional neural network to realize enhancement of the low-illumination image, constructs a loss function to perform constraint optimization on parameters of the convolutional neural network, so as to achieve the expected effects of low-illumination image brightness, contrast enhancement and image impression enhancement, remove the influence of noise and color distortion to a great extent, and allow a user to autonomously adjust and enhance the brightness.

Description

technical field [0001] The invention relates to the technical field of digital image processing, in particular to a low-light color image enhancement method based on Retinex and convolutional neural network. Background technique [0002] With the popularization of digital products, especially smart phones, people can conveniently collect various image information. In real life, many images are taken under low light or unbalanced light conditions. These images often have low visual quality problems, such as darkening of the whole or some areas of the image, difficulty in capturing details, color distortion, and noise. Seriously wait. These problems of low-light images seriously affect people's visual experience or computer processing of images. Low-light image enhancement technology can enhance low-light images, thereby adjusting the brightness of the image, restoring the details of darker areas in the image, and helping people or computers to perform further image analysis...

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
IPC IPC(8): G06T5/00G06N3/04G06N3/08
CPCG06N3/082G06T2207/10024G06N3/045G06T5/92
Inventor 张永华郭晓杰张加万
Owner TIANJIN UNIV
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