Research On Enhancement Algorithms Of Low Illumination Images Based On Human Visual Characteristics

In real life,an image or a video acquired by a capture device under low illumination environment may suffer from the problems of contrast dropped,detail loss,color distortion.These problems seriously affect the subsequent processing and application.Therefore,it is important to enhance the degraded image in such environment.This thesis analyzes the reasons for image degradation and characteristics of a degraded image capture in low illumination environment,investigate a number of common image enhancement algorithms and makes improvement of existing methods.The main points of the research are as follows.Firstly,a new enhancement algorithm based on logarithmic image process(LIP)is proposed to solve a single degraded image.In order to improve the extraction effect of reflection components,we use the improved filter based on LIP model to obtain the reflection image,so as to keep high frequency components better.We also propose illumination components enhancement algorithm based on local background brightness according to Weber-Fechner law,which can improve the brightness and contrast of an image and avoid the over enhancement in local regions.Experimental results show that compared with traditional enhancement algorithms,the proposed algorithm based on human visual properties can improve the overall contrast of an image effectively,preserve the details well,and have better subjective visual perception.Secondly,an improved video enhancement algorithm based on image fusion is proposed.Traditional algorithms have the problems of image aliasing,color drift and foreground weak enhancement of motion region.To solve these problems,an improved fusion enhancement algorithm based on information compensation of high quality video frames is proposed.In our algorithm,we separated the foreground and background of the static background firstly,then the background enhanced by an improved fusion strategy and the target of the previous scene enhanced by contrast limited adaptive histogram equalization.Finally,the background and the moving object are fused to get the last enhancement image.Experimental results show that compared with some traditional algorithms,the improved algorithm can highlight image details better,improve the contrast in bright and dark regions,and solve the problem of confusion and color drift.Finally,a comprehensive evaluation for image enhancement is implemented.Effective and unified quality evaluation method of image enhancement is lacked at present,so we propose an evaluation method based on human visual characteristics and establish the corresponding evaluation function.Experimental results show that this method can offer a more accurate evaluation of the enhancement effect on a degraded image in insufficient illumination environment.