强光源影响下的单幅图像去雾  

Single Image Dehazing with the Influence of Intense Light Source

作　　者：毕雪英 过洁[1] 潘金贵[1] 

机构地区：[1]南京大学计算机科学与技术系,江苏南京210023

出　　处：《常州大学学报（自然科学版）》2015年第1期86-92,共7页Journal of Changzhou University：Natural Science Edition

摘　　要：针对强光源影响所造成的非均匀大气情况,通过分析强光源对均匀大气雾天图像退化模型的影响,将大气光分量分为全局大气光常值分量和强光源影响分量,从而提出了一种针对强光源影响下的雾天图像去雾新方法。该方法首先自动获得图像中强光源的中心位置及拟合区间,然后利用最小二乘法计算强光源影响分量的二维高斯拟合,在输入雾图中去除强光源影响分量后得到大气光常值分量的估计值,进而根据暗通道先验知识估算大气透射率,最后利用所得参量对雾天图像进行去雾。实验结果表明,该去雾方法对户外强光源影响下的自然雾天图像复原效果较好。In view of the inhomogeneous condition influenced by the intense light source,a novel single im- age dehazing approach is proposed for images containing intensively emitted light source. This approach builds upon an image degradation model introduced via physically analyzing the illuminating effects of in- tense light source. In this paper, based on the experimental and observational discovery, the atmospheric light is divided into global constant component and intense light component. Firstly,the location and fitting range of light source is obtained. Then,the 2-D Gaussian fitting function of intense light component is cal- culated by using least square method,and the global constant component can be estimated after removing the strong light influence part in the input image. Thereafter, the atmospheric transmittance is estimated according to the dark channel prior knowledge on the light source removed image. Finally, the hazy image can be restored by the inverse process of the degradation model. The experimental results show that the new algorithm is effective for actual outdoor degraded images.

关 键 词：强光源影响 去雾 退化模型 最小二乘法 二维高斯拟合 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]