摘要
航向重叠度小于53%,不满足连续3张影像进行模型连接的航空影像视为非常规航空影像。针对非常规航空影像与机载激光扫描(Light Detection And Ranging,LiDAR)数据的配准,本文提出了首先利用影像匹配得到匹配点云,然后基于迭代最邻近点(Iterative Closest Point,ICP)算法配准匹配点云和机载LiDAR点云,最后使用配准点进行单像空间后方交会解算影像方位元素的配准方法。相关实验表明利用本文方法配准航空影像和机载LiDAR数据,相对于人工配准,自动化程度大大提高,精度更高。
The aerial images which forward overlap is less than 53%,and do not satisfy model connection conditions using continuous three images are regarded as unconventional aerial images.For registering unconventional aerial images with airborne LiDAR (Light Detetion And Ranging)data,this paper presents a new registration method that using image dense matching to get matching point cloud firstly,and then register matching point cloud to LiDAR points by using ICP (Iterative Closest Point) algorithm,finally calculate orientation elements of images by single image spatial resection with registration points. Experimental results indicated that this proposed method has a great advantage in higher automation and precision compared with the manual registration.
出处
《遥感信息》
CSCD
2014年第6期16-20,共5页
Remote Sensing Information
基金
中国南方电网公司重点科技无人机电力线路巡检智能诊断系统研发项目(K-GD2013-030)
国家测绘地理信息局基础测绘项目(A1414)
关键词
非常规航空影像
机载LiDAR数据
影像密集匹配
ICP算法
配准
单像空间后方交会
unconventional aerial images
airborne LiDAR data
image dense matching
Iterative Closest Point algorithm
registration
single image spatial resection
