摘要
针对雾霾条件下拍摄的户外图像,常规去雾后天空区域常常出现的失真问题,提出了一种结合天空区域检测的图像去雾算法;算法先根据暗通道理论估计出大气光强度,使用双边滤波器得到大气光幕,求得透射率图,再结合天空区域检测的结果对透射率进行修正,最后代入雾天成像模型得到复原的图像;实验结果表明:结合天空区域检测的图像去雾算法可以有效地检测出图像中是否存在天空区域,针对检测结果修正的透视率,能够使修复后有天空区域的图像看起来更加自然平滑,没有明显失真,不存在天空区域的图像,图像对比度大大提升,在景深较大的区域恢复出更多的细节;算法对各类图像均可取得较为理想的去雾效果。
According to outdoor images taken under hazy condition,sky area after dehazing is often distorted,based on that,this paper proposes a single image dehazing algorithm combined with sky region detection. Firstly,the atmospheric light intensity is estimated based on the dark channel theory,bilateral filter is used to obtain the atmospheric light and transmissivity map,then sky detection procedure is applied to find whether the sky area is existed in image,the transmissivity is modified according to different results,and finally the restored image is obtained by substituting the fog imaging model. The experimental results show that the proposed algorithm can effectively detect the sky area in image,can modify transmissivity according to modified results and can make the image in the sky region look more natural and smooth and no obvious distortion. It can also improve the contrast of the image without the sky areas which recover more details in the depth of field. This algorithm can receive ideal dehazing effect for all kinds of images.
出处
《重庆工商大学学报（自然科学版）》
2017年第5期37-42,共6页
Journal of Chongqing Technology and Business University:Natural Science Edition
基金
国家自然科学基金(61372068)
广东省农村信息化建设专项资金资助项目(201210112700518)
关键词
图像去雾
双边滤波
暗通道
天空检测
区域生长
image dehazing
bilateral filtering
dark channel
sky detection
region growth
