序列无人机影像拼接方法研究  

Sequence UAV image mosaic method

作　　者：唐敏[1] 杨锋[1] 黄华平[1] 

机构地区：[1]中国中铁二院集团有限责任公司,四川成都610031

出　　处：《测绘工程》2016年第1期10-16,共7页Engineering of Surveying and Mapping

摘　　要：无人机影像具有数量多、畸变大、POS数据不准确等特点,导致其在拼接过程中会产生大量的累积误差,要快速地获得大范围准确的全景图有一定的困难。基于此,提出一种既精确又高效的无人机序列影像拼接方法。首先计算大致的影像匹配区域,减少特征搜索和匹配的时间,同时记录匹配区域中心位置的特征点坐标,引入平差理论,区分平地、丘陵、山区等不同区域加权纠正匹配特征点的坐标位置。同时针对航带间重叠率小、姿态差异大等特点,采用"先航带间再航带内、旁边航带向中间航带靠拢"的拼接方式,减少整体区域的累积误差产生,最终完成全局影像的拼接。UAV images have large number of quantities,big distortion and other characteristic,which makes a large amount of accumulated errors in the splicing process.The access to a wide range of accurate panorama will face certain difficulties.Based on this,this paper presents an accurate and efficient method of UAV image sequence stitching.Firstly,it calculates the approximate area of image matching,reduces the time of feature search and matching,and at the same time,records the coordinates of feature point,which is at the location of regional centers.

关 键 词：无人机影像 拼接 航带 累积误差 分区域加权 

分 类 号：P231[天文地球—摄影测量与遥感]