城区机载激光雷达点云数据与航空影像的多尺度配准方法

Multi-Scale Registration of LiDAR Data and Aerial Image

提出了一种激光雷达(LiDAR)点云数据和航空影像的多尺度配准方法.该方法采用渐进式的配准策略,利用面形态学的方法构建尺度空间.配准过程分为两部分:在大尺度下利用面片特征进行配准;在小尺度下利用直线特征进行配准.渐进式的配准策略简化搜索匹配的直线特征的过程,提高了自动化程度,构建基于面形态学的尺度空间提高了配准元的提取精度.最后,通过实验验证了方法的可行性.

A multi-scale registration method was presented for aerial image and light detection and ranging（LiDAR）data.Multi-scale registration refers to a＂from coarse to fine＂and progressive registration strategy based on scale-space theory.Area morphology was used to build the scalespace.The methodology includes two parts：first,registration in coarse scale with surface patch and second,registration in fine scale with linear feature.In this proposed method,the search for matched lines was simplified as a result of progressive registration strategy.The automaticity was improved.Besides,the area morphological scale-space was used to improve the accuracy of extraction of the registration primitives.This method was verified with experiment.