基于均值漂移的暗原色先验图像去雾算法  

An image haze removal algorithm using dark channel prior based on mean shift

作　　者：陆海俊 汪荣贵[1] 杨娟[1] 薛丽霞[1] 王晶[1] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2016年第9期1205-1210,共6页Journal of Hefei University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(61075032);中央高校基本科研业务费专项资金资助项目(2012HGCX0001);合肥工业大学博士学位人员专项基金资助项目(JZ2014HGBZ0059)

摘　　要：暗原色先验去雾算法在单幅图像去雾方面效果明显,但该算法复杂度高、处理耗时长,针对该算法不足之处,文章对均值漂移算法进行适当改进后,将其引入暗原色先验去雾算法之中,并结合数学形态学的思想,提出一种新的高精度透射率快速计算技术和相应的去雾算法。通过改进的算法,有效解决了块状效应和光晕现象的问题。选取一系列雾天图像进行实验,分别采用主观观察评价和客观数据分析方法,结果表明文中所提出算法快速有效,并较好地恢复了图像细节,减少了色彩失真,同时改进算法处理效果与原算法基本一致,算法效率得到显著提高。The image haze removal algorithm based on the dark channel prior has achieved a good result in removing the haze from foggy images ,but it has high computational complexity of transmission and is time consuming .In this paper ,the mean shift algorithm is improved properly and introduced into the image haze removal algorithm using the dark channel prior based on the ideas of mathematical morphology .Then a new kind of high precision transmission estimation algorithm is put forward .The block effect and halo effect can be eliminated effectively by using this method .A series of foggy ima‐ges are tested ,and by using the methods of subjective observation and objective data evaluation re‐spectively ,the proposed algorithm is proved to be quick and effective ,and is better in restoring image details and reducing the color distortion at the same time .The proposed algorithm has almost the same processing effect as the original algorithm and higher efficiency .

关 键 词：图像去雾 暗原色先验 透射率估计 均值漂移 图像质量 形态学 

分 类 号：TP391.411[自动化与计算机技术—计算机应用技术]