改进暗通道先验的高分二号影像快速去薄云  

Fast Removal of Thin Cloud from GF-2 Satellite Imagery Based on Improved Dark Channel Prior Algorithm

作　　者：潘勇卓 谢洪斌[1,2] 杨雪[1,2] 席亚文 戚伟迅 PAN Yongzhuo;XIE Hongbin;YANG Xue;XI Yawen;QI Weixun(Chongqing Key Laboratory of Exogenic Mineralization and Mine Environment,Chongqing Institute of Geology and Mineral Resources,Chongqing 400042,China;Chongqing Research Center of State Key Laboratory of Coal Resources and Safe Mining,Chongqing 400042,China)

机构地区：[1]重庆地质矿产研究院外生成矿与矿山环境重庆市重点实验室,重庆400042 [2]煤炭资源与安全开采国家重点实验室重庆研究中心,重庆400042

出　　处：《遥感信息》2019年第6期91-98,共8页Remote Sensing Information

基　　金：重庆市国土房管科技项目(KJ-2018011)

摘　　要：为了有效消除薄云对高分二号影像的干扰,提出了一种基于改进暗通道先验的遥感影像快速去薄云方法。首先设置最大大气光阈值,解决原始暗通道先验法对高亮区域的过校问题;然后使用下采样和插值算法简化透射率图的输入,大幅提高处理速度;最后针对成果图像的偏蓝现象,利用散射模型以蓝光波段推算其余波段的改正量。实验结果表明,该方法能有效去除高分二号影像中的薄云,且在纹理色彩、校正真实性和处理效率上都有较好的表现,具有较高的实用价值。Based on dark channel prior(DCP)algorithm,an improved fast method is proposed in this paper to remove thin cloud from single frame GF-2 satellite image.The obvious flaws of DCP algorithm are over-correction in high luminance region and blue shift in final images.For improvement,following steps based on DCP are implemented.Firstly,the maximum global atmospheric light threshold value is defined to solve the problem of image over-correction in high luminance region.Secondly,blue band is used as initial data to calculate the offset gray scale value of other bands under atmospheric scattering model,which can eliminate the appearance of blue shift effectively.Finally,processing speed is another important aspect to consider other than quality.In the processing of transmission rate calculation,the input image is simplified using the resampling algorithm and interpolation algorithm in order to improve speed of operation.Experimental results show that the proposed method can lead to better cloud removal quality such as satisfying detailed texture,better color accuracy and reliable rectification.Processing speed is also improved significantly.This method is valuable particularly for single frame GF-2 satellite image cloud removal.

关 键 词：薄云去除 高分二号 暗通道先验 散射模型 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]