一种雾天退化场景快速复原方法  

A Fast Method for Haze Image Restoration

作　　者：王伟鹏[1] 戴声奎[1] 项文杰[1] 

机构地区：[1]华侨大学信息科学与工程学院,福建厦门361021

出　　处：《华侨大学学报（自然科学版）》2015年第2期156-160,共5页Journal of Huaqiao University(Natural Science)

基　　金：福建省科技计划重点项目(2013H0030);中央高校基本科研业务费专项基金资助项目(JB-ZR1145)

摘　　要：针对雾天条件下图像退化的问题,提出一种新的场景复原方法.首先,结合大气散射模型与光学反射成像的特性,对大气耗散函数提出物理约束条件,利用双边滤波的边缘保持特性得到大气耗散函数的初始估计.然后,引入局部均值与局部标准差近似估计图像对比度,从而降低对比度较强的部分区域的雾浓度估计来获取更加准确的大气耗散函数.最后,求解雾图成像方程,恢复理想光照条件下的辐射强度.实验结果表明:该方法对图像的细节和颜色恢复非常有效,并能有效克服边缘残雾现象以及颜色过饱和等不足.According to the problem of image degraded in hazy condition,a new haze image restoration method is proposed.Based on the physical property of atmospheric scattering model and optical reflectance imaging,the physical constraint of atmospheric veil is presented at first.The rough atmospheric veil is obtained by using bilateral filter which has good abilities of keeping edge.Then the local mean and local standard deviation are introduced to estimate the contrast of image,therefore the estimation of local haze density is decreased through good contrast region in order to get more accurate atmospheric veil.Finally,the ideal scene radiance is recovered through solving the equation of haze image.Experimental results show that this method not only achieve great restoration for image detail and color information,but also avoid the shortcomings of edge effect and over saturated.

关 键 词：去雾 图像复原 大气散射模型 双边滤波 图像增强 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]