基于局部一致性的马尔可夫随机场去雾

为克服暗通道先验的适用局限性,同时增强一阶马尔可夫随机场对图像全局信息的约束能力,在颜色衰减先验的基础上,提出了一种局部一致马尔可夫随机场(Markov random fields, MRF)单幅图像去雾算法。首先,结合颜色衰减和暗通道两先验假设的特征，获取普适性更强的介质传输图粗估计,然后利用基于颜色特征的图像局部一致块代替MRF的二阶及其高阶能量项来构造代价函数,达到优化介质传输图和获取最终去雾图像的目的。实验结果表明,所提算法可以获取细节保持更好且鲁棒性更强的去雾效果。

Markov random fields defogging based on local consistency

To overcome the limitation of dark channel prior’s application, and strengthen the first-order Markov random fields (MRF) constraint ability of the global image information, a local consistent MRF defogging method is proposed based on color attenuation. First, combining with the advantages of color attenuation and dark channel priors, a more robust estimation of medium transmission is obtained. Then, the cost function is constructed with the color features based consistent blocks instead of Markov random fields’ two-order and higher-order energy terms. Finally, the defogged image is obtained.The experimental results show that this method could improve the image resolution.