Research On Enhancement Algorithm For Low Illumination Image Based On LIP Model

Image processing technology is playing an important role on human access to information with the development of technology,among which the low illumination image enhancement is related to public security,night traffic,criminal detection and so on related to people’s livelihood.There are several problems in low illumination image such as the nonuniform illumination,the low brightness and the image detail loss.Therefore it has always been a hotspot and task in image enhancement.This paper firstly introduces some traditional image enhancement theories including color space,spatial and frequency domain theory and the Retinex,among them it analyses the characteristics and defects of several classic Retinex algorithms by experiment.After that the chapter introduces the classic logarithmic image processing(LIP)model and other extented models according to different demand in order to solve the over-interval problem in linear processing and to better correspond the human visual characteristic.Moreover,the chapter applies the LIP to image enhancement and edge detection.Secondly,in the view of the color graying and distortion of MSRCR and the nonuniform illumination problem,an improved MSRCR algorithm for illumination division in LIP model is proposed in this chapter.The algorithm better reflects the advantages of different scales Retinex algorithm by using appropriate scale operation in different illumination area and the improved color recovery factor.Finally,since the original Lee algorithm can not well adapt to the nonuniform illumination,as well as the traditional LIP algorithms cannot deal with negative part of image,this chapter puts forward a new adaptive parameters Lee algorithm based on the PSLIP.The algorithm can not only operate in two mode of reflected and transmitted light,but also can better balance the results of different lighting conditions and complexity using the adaptive parameters.In addition,it use the bilateral filtering with improved weight coefficient in order to eliminate the influence of noise.