基于统计直方图的单幅图像去雾算法  

SINGLE IMAGE DEFOGGING ALGORITHM BASED ON STATISTICAL HISTOGRAM

作　　者：董蕊[1,2] 王志巍[1,3] 刘淑娟[1,3] 张有会[1,3] 赵晔[1,3] 

机构地区：[1]河北师范大学数学与信息科学学院,河北石家庄050024 [2]河北交通职业技术学院,河北石家庄050091 [3]河北省计算数学与应用重点实验室,河北石家庄050024

出　　处：《计算机应用与软件》2016年第5期196-199,251,共5页Computer Applications and Software

基　　金：河北省高等学校科学技术研究项目(Z2013061)

摘　　要：针对雾天条件下图像的对比度及色彩的饱和度等不同程度的退化现象,提出一种简单可行的单幅图像去雾算法。首先,对带雾图像的暗通道图作统计直方图分析;其次,结合后谷查找法、后峰查找法,有效划定天空区域范围;再次利用统计结果中天空部分比例最高的亮度参数,自适应地确定大气光线强度;最后利用所得数据结合雾天图像物理模型作去雾处理。实验结果表明,该算法能够有效地、自适应地判定天空范围,清晰地恢复原始图像细节,使去雾后的图像更逼近原始色彩。Under foggy conditions,both the contrast and colour saturation of an image will be degraded. To address this issue,we proposed a simple but feasible defogging algorithm for single image. First,we made statistical histogram analysis on dark channel of the image with fog. Secondly we scoped the range of the sky effectively in combination with last valley search method and last peak search method;Then,we adaptively determined the intensity of atmospheric light by employing once again the brightness parameter which was the highest proportion of the sky in statistical histogram. Finally,we carried out defogging processing with the derived data and combining the physical model of foggy days. Experimental results showed that the algorithm could determine the scope of the sky effectively and adaptively,and restored the details of original image clearly,thus made the defogged image be closer to original colour.

关 键 词：图像处理 统计直方图 天空范围判定 后谷查找法 后峰查找法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]