基于YUV色彩空间的暗原色单幅图像去雾改进算法  

An improved approach on single image dehazing using modified dark channel prior in YUV color space

作　　者：张鹏[1] 陶青川[1] 黄凤英 

机构地区：[1]四川大学电子信息学院 [2]78009部队

出　　处：《太赫兹科学与电子信息学报》2016年第2期287-292,共6页Journal of Terahertz Science and Electronic Information Technology

基　　金：国家自然科学基金资助项目(61471248)

摘　　要：为得到更清晰的复原图像,提出了一种基于暗原色先验理论图像去雾的改进算法。先将原始雾天RGB图像转化到YUV色彩空间,在Y(Luminance)通道中,对像素点进行区域划分,只使用属于同一区域的像素点来计算该方块的暗原色,同时在Y通道计算大气光值;最后,通过修正复原图像的计算公式,来校正亮区域色彩失真。与基于引导滤波及RGB色彩空间的暗原色去雾算法相比,本文算法具有更好的去雾效果。An improved approach on single image dehazing is proposed in order to obtain clearer restored images. Firstly, the source images are converted into YUV color space, and then, pixels are divided into different areas, only those pixels belong to the same areas in Y channel are used to calculate Dark Channel, and the air light value are also calculated in Y Channel. Finally, the distortion of bright area is adjusted through modifying the formula by which the restored image is calculated. Compared with those algorithms based on dark channel prior with guided filter in RGB color space, this algorithm is more effective.

关 键 词：图像去雾 暗原色先验 分段暗原色 亮区域校正 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]