摘要: |
针对在雾霾环境下获取的图像降质严重、现有算法去雾图结构细节信息丢失较多的问题,提出了一种结合暗通道先验(DCP)和马尔可夫随机场(MRF)的单幅图像去雾算法。该算法先采用子块部分重叠局部直方图均衡(POSHE)对原始雾图进行增强,以提高其对比度,并通过DCP算法获取优化后的透射率;利用MRF模型对图像结构细节信息的约束特性,对透射率进行建模,以进一步细化透射率;由天空域的显著特征,通过分块搜索法求取大气光值。与传统去雾算法相比,该算法能得到更精确的透射率图,有效保持图像结构信息,去雾后的图呈现出丰富的细节和较真实的色彩视觉效果。 |
关键词: 图像去雾 暗通道先验 马尔可夫随机场 分块搜索法 直方图均衡 |
DOI: |
|
基金项目:重庆市研究生科研创新基金资助项目(CYS15166) |
|
A single image dehazing algorithm based on dark channel prior and MRF theory |
WU Cuixian,ZUO Xing |
(Research Center of New Telecommunication Technology Applications,
Chongqing University of Posts and Telecommunications,Chongqing 400065,China;
Chongqing Information Technology Designing Co.,Ltd.,Chongqing 400065,China) |
Abstract: |
Aiming at the problems that the quality of image obtained in the haze environment is seriously degraded and existing algorithms can't preserve the details of image,this paper proposes an image dehazing algorithm based on Dark Channel Prior(DCP) and Markov Random Field(MRF) theory.Firstly,the Partially Overlapped Sub-block Histogram(POSHE) algorithm is used to improve the contrast of the original fog image.Meanwhile,the optimized transmission map is obtained by DCP algorithm.Then,the constraint characteristics of MRF model are adopted to model the transmission map for refining it.Finally,the block search method is used to get the atmospheric light value based on the characteristics of the sky domain.Compared with traditional methods,the proposed method can obtain more accurate transmission map and effectively preserve the structure information of image.The restored image presents rich details and realistic color visual effects. |
Key words: image dehazing dark channel prior(DCP) Markov random field(MRF) block search method histogram equalization |