Retinex Image Enhancement Based On Particle Swarm Optimization

People combine the optimized digital imaging equipment with rapid development of computer technology, The image enhancement technology play an important role on improving the quality of photos, becoming one of the focus in researching image processing. With global features, the Retinex theory is a classical algorithm of image quality enhancement basing on the color conatancy of human visual system, getting the essential features of goals by simulating observation mode of eye, This algorithm also has some advantages of significant dynamic compression, color constancy, tone reproduction and so on. In this report, we learned and analysis the classic single-scale Retinex image enhancement algorithm, found the deficiencies, besides, we designed a Retinex image enhancement improved method basing on the optimized particle swarm, getting better effection of enhancing. The main research work as followed:1)To learn and research the concepts and methods of image enhancement processing, we also analysis the current situation of Retinex image enhancement processing, which has single-scale algorithm and multi-scale algorithm, this article will focus on deep analysis of single scaling algorithm and designing improved algorithm.2) Because the traditional single-scale Retinex image enhancement algorithm has disadvantage of all pixels using the same spatial filter template, we design a modified single-scale Retinex image algorithm which is combined with the fast particle swarm optimization and Retinex image algorithm. The algorithm is to convert the image to HIS image color space, clssifying the pixel according to differents values of H, different types of pixels use different filter templates, we select a specific filter template by PSO which is decided by values of entropy size expecting to get better image enhancement;3)Under environment of MATLBA, we did the simulation contract test between the improved algorithm and the traditional single-scale Retinex color image contrast enhancement algorithm, comparing enhanced lighting effection of three pictures with different scenes and images, from the visual point of view, this Retinex image enhancement algorithm based on PSO achieved better results including higher values and richer information.