Low Illumination Color Image Enhancement Alorithm Research Based On Retinex Theory

Color images have richer color information compared with other types of images,However,in the process of collecting color images,the images under low-light conditions can not meet the requirements index of the application during the night when the lighting conditions are not sufficient,which brings a lot of trouble on night-time video monitoring work.Due to the above illumination conditions,it's unable to get high quality color images,this situation seriously affects the use value of the image.Therefore,the enhancement of low-illumination color image has very important practical significance.The specific work of this subject includes the following points:1.This article illustrated the Retinex algorithm and its mathematical model,explained the core theory of the algorithm,illustrated the impact of the scale parameter value on the image enhancement effect.2.This paper studied the Retinex image enhancement algorithm of improvement reconstruction adaptive weights,in HSI color space model,the saturation component is stretched in a linear way,reflection component is calculated which is used by guided filtering multi-scale Retinex algorithm of improvement reconstruction adaptive weights,Gamma correction method for reflection component obtain enhanced brightness component,finally the low-illumination color image is enhanced using by color restoration.The algorithm has the advantages of significant enhancement effect,short running time,rich image detail and color distortion avoidance.3.This article researched the luminance-stratified Retinex image enhancement algorithm based on the guided filtering,in HSI color space model,the luminance image is filtered using by the leading filter combined with the multi-scale Retinex algorithm,and the luminance image is divided into the detail image and the basic image with the image stratification technique,the detail image and the basic image are respectively enhanced and recombined to obtain the enhanced brightness image with the gain coefficient,and the color restoration operation is performed in the RGB color space.The algorithm can make the edges and details of the image more prominent and avoid the "halo artifact" situation.Image quality assessment methods are used to evaluate the enhancement results of the improved algorithm.The proposed improved algorithm can significantly improve the image quality of low-light color images and can be adaptively enhanced according to the light environment of the images,there is important theory research value and application prospect.