期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2010
卷号：XXXVIII - Part 1
出版社：Copernicus Publications
摘要：In close range photogrammetry or laser scanning, it is often not possible to image or scan certain objects of interest or enclosed spaces from a single sensor station. In some cases, it is necessary to produce multiple irregular point clouds or surface models from different sensor locations, i.e. one unique point cloud from each sensor location. These separate point clouds or surface models belong to different coordinate systems, and in order to fuse all points in a single dataset, the surface models have to be registered to a common reference frame. This paper describes a methodology for performing registration of such multiple surface models. First, conjugate point-patch pairs are detected in the overlapping surface areas, and the transformation parameters between all neighbouring surfaces are estimated in a pairwise manner. Then, using the conjugate point-patch pairs, and applying the transformation parameters from the pairwise registration as initial approximations, the final surface transformation parameters are solved for simultaneously. This is done in a least-squares adjustment, where each surface is iteratively transformed to a common reference frame until the sum of the squared normal distances between the conjugate surface elements is minimized. This paper will show two ways of performing this least-squares adjustment. One is referred to as the coplanarity constraint method, and the other one as the modified weight matrix method. The paper will compare results for the multiple-surface registration of an artificial scoliotic torso mannequin using both approaches
关键词：close range; surface; registration; point cloud; comparison; analysis; method
