点云中角点的自动提取算法研究  

Automatic corner detecting algorithms of point cloud

作　　者：孟敏[1] 李嘉[1] 陈小浩[1] 田建[1] 

机构地区：[1]河海大学测绘科学与工程系,江苏南京210098

出　　处：《测绘工程》2013年第1期31-34,共4页Engineering of Surveying and Mapping

基　　金：中央高教基本科研业务费专项资金资助(2009B02814);江苏省交通科学研究计划项目(09Y08)

摘　　要：针对三维激光扫描数据处理过程中的数据拼接、纹理映射、目标识别等问题,提出一种新的三维激光扫描距离图像的角点提取算法,旨在为相邻测站间数据拼接提供粗配准的对应点。文中通过把三维激光扫描仪获得的单站距离图像转化为二维深度图像,利用已有的SUSAN算子提取出深度图像的角点,反算深度图像的角点到距离图像坐标系下实现角点的提取。利用2站间4个公共角点作为粗配准进行2站数据的ICP配准,利用配准的准确度验证该角点提取算法的有效性。According to the rising problems in the process of data processing such as data registration, texture mapping and target identification, it presents an algorithm of corner detecting, aiming at easily founding the corresponding points between adjacent stations. First, make each station＇s three-dimensional distance image into two-dimensional gray image, using the relative position between the instrument center and scanned target; second, use SUSAN operator to detect the corner points in the two dimensional gray image; last, match the corner points that have been detected, using the stitching precision to prove the effectiveness of this algorithm.

关 键 词：距离图像 深度图像 SUSAN算子 角点提取 

分 类 号：P237[天文地球—摄影测量与遥感]