Model Based On Visual Characteristics Of The Image Enhancement Algorithms With Performance Analysis

Image enhancement is one of the basic problems in the low layer of image vision processing. How to combine the sensing characteristics of human vision system(HVS) to the contrast\color\saturation in the image to design algorithms about image enhancement with color fidelity and obvious details is one of the hotspots in the field of image enhancement. In this dissertation, the improvement of some kind of algorithms which are based on the Retinex theory is chosen as the research idea, with the basic overview of the research status about image enhancement algorithms, with the main research line about the color image enhancement, the main tasks in this dissertation concludes:First of all, combining the global tone mapping algorithm and the histogram rescaling algorithm, design and realize a improvement of color image enhancement which on the basic of Retinex in the domain of multi-scale Cortex transform. The experiments on the algorithm shows that it can improve the visual effect about the original images, recover the detail information which are hidden in the dark zone, maintained the hue in the original image without adding any additional light resources, halo artifacts and the block effect in the dark zone.Secondly, according to the first order statistics information, local mean and standard deviation in the original image, design the reference value in the local window tone mapping model, implement the local tone mapping color image enhancement algorithm. The experiment on the algorithm shows that it can improve the contrast information and maintain the local structure in the image, recover the detail information in the dark zone with decreasing the halo artifacts.Last, based on the sensing characteristics about the HVS to the natural image, an image contrast enhancement algorithm which is according to the optimized energy functional to increase the contrast and maintain the chrominance is researched. In the dissertation, a color image contrast enhancement algorithm is designed and implemented which is based on the iteration to attain the minimal solution about the energy functional. The experiment on the algorithm shows that it can improve the contrast information in the image while maintain the hue information and improve the saturation in the image.