一种改进的Retinex彩色图像增强方法  

Color Image Enhancement Based on Improved Retinex Algorithm

作　　者：王小鹏[1] 陈璐[1] 魏冲冲[1] 张雯[1] 

机构地区：[1]兰州交通大学电子与信息工程学院,甘肃兰州730070

出　　处：《兰州交通大学学报》2015年第1期55-59,70,共6页Journal of Lanzhou Jiaotong University

基　　金：国家自然科学基金(61261029);兰州市科技计划项目(2013-4-63);金川公司预研基金(JCYY2013009)

摘　　要：传统Retinex算法在对彩色图像进行增强时,容易产生光晕以及处理后图像整体对比度较低等问题.为此提出了一种改进Retinex的图像增强方法,该方法采用三高斯模型和高斯滤波相结合的双边滤波对照度分量进行估计;然后在提取反射分量时,引入原始图像亮度影响因子实现区域自适应增强;最后进行归一化调整和彩色图像恢复,实现图像增强.实验结果表明,该方法能够在增强图像的同时,有效抑制光晕,提高图像的整体对比度.Traditional Retinex algorithm is prone to the problems of halo and lower holistic contrast,so an improved Retinex algorithm is proposed.Firstly,the bilateral filtering which combines tri-Gaussian model filter and Gauss filter is used to estimate the illumination components,then the original image brightness factor is taken into account to realize the regional adaptive enhancement during the reflection components extraction.Finally,the image enhancement is implemented by the normalized adjustment and color image restoration.Experiments show that the method can efficiently enhance the image,inhibit the halation and improve the holistic contrast.

关 键 词：图像增强 RETINEX算法 三高斯模型 双边滤波 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]