基于改进Retinex方法的遥感图像云雾去除研究  

Research on cloud elimination of remote sensing image based on improved Retinex method

作　　者：刘鑫[1] 祝海江[1] 宋光磊[2] 

机构地区：[1]北京化工大学信息科学与技术学院,北京100029 [2]中国航天科技集团公司中国空间技术研究院五一三所,北京100086

出　　处：《北京化工大学学报（自然科学版）》2013年第B12期76-80,共5页Journal of Beijing University of Chemical Technology(Natural Science Edition)

摘　　要：针对目前遥感图像受云雾覆盖导致利用率较低的问题,通过分析Retinex理论在云雾去除上的不足,提出了一种基于图像云雾区域识别的Retinex云雾去除改进新方法。通过划分出遥感图像的云雾覆盖区域和无云区域,并对不同的区域有针对性地进行处理,达到同时进行图像云雾去除和地面信息增强的效果。对仿真实验结果的分析表明,此种新方法能够在图像云雾去除、色彩恢复及细节增强等方面提供更好的处理效果,大大提高了遥感图像的可读性,能够有效提高遥感数据的利用率,为图像的后续处理做准备。Aiming at the problem that the utilization coefficient of the remote sensing image is comparatively too low,this paper proposed a new method of improved Retinex clouds elimination based on image cloud region recognition.By dividing the remote sensing image into cloud covered region and non-cloud region and processing different regions with pertinence,the image cloud was eliminated and the ground information was enhanced.Analysis of the simulation experiment showed that this method could provide better effect on image cloud elimination,color recover and details enhancement etc.This method could greatly improve the readability of remote sensing image and the utilization coefficient of remote sensing data,making preparations for follow-up image handling.

关 键 词：RETINEX理论 亮度保持因子 云雾去除 区域识别 

分 类 号：V557.3[航空宇航科学与技术—人机与环境工程]