摘要
天绘一号卫星2m分辨率全色相机搭载8片TDICCD(时间延迟积分耦合器)。为获取大宽幅的影像,后期处理中需将8个条带影像进行拼接。在基于SIFT算法进行相邻两片CCD特征点匹配的过程中,如果在重叠区内有云层存在,云层周围的特征点会随着云的移动而移动,从而影响拼接精度。本文重点分析了云动对拼接精度的影响,并对SIFT算法进行了改进。试验结果表明,该算法可以有效消除云动对影像拼接精度的影响。
The camera of two meter resolution panchromatic camera of TH-1 satellite equits with eight TDICCD(Time Delay Integration Coupler).To obtain a large width of the image,post-processing need to eight bands image stitching.During the processing of SIFT algorithm based on two adjacent CCD feature point matching,if the presence of the overlap region with clouds,clouds around the feature points will move with the movement of the clouds,thus it will affect the splicing accuracy.This paper focused on the moving clouds on stitching accuracy,and improved SIFT algorithm.Test results show that the algorithm can effectively eliminate moving clouds which affect stitching accuracy of the image.
出处
《遥感信息》
CSCD
2014年第3期57-61,共5页
Remote Sensing Information
