差分分层暗原色先验去雾法  

Differential layering Haze-removal method based on the prior dark-channel

作　　者：李宁[1] 洪德成[2] 

机构地区：[1]石家庄学院物理与电气信息工程学院,石家庄050035 [2]吉林大学物理学院,吉林长春130012

出　　处：《重庆大学学报（自然科学版）》2015年第3期136-141,共6页Journal of Chongqing University

基　　金：国家自然科学基金资助项目(41304084)~~

摘　　要：针对烟尘和雾霾等恶劣户外环境下视频图像严重退化的问题,结合大气散射模型和暗原色先验去雾的原理,针对原始算法中暗区域扩张的缺点,提出一种区分图像景深边界的方法。此方法通过对像素求差分能够将不同景深的像素点区分,进而分别求出不同景深图像的透射率,从而达到快速、准确地恢复图像的目的。实验结果表明,这种方法能够具有更高的处理速度和图像细节,更接近实时去雾的要求。In the smoky foggy hazy and other harsh outdoor conditions,scene images captured severely suffer from poor contrast and visibility.Aiming at the problem,an improved method is proposed to overcome the dark region expansion quickly on dark-channel prior.By calculating difference of dark channel,the method can effectively distinguish the different depth of fields in adjacent area approximately and estimate transmissivity.Experimental results demonstrate that the new method provides more speed and higher dehazing quality than other techniques.

关 键 词：去雾 暗原色 复原 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]