AUTOMATIC REGISTRATION OF MULTIPLE LASER SCANS USING PANORAMIC RGB AND INTENSITY IMAGES

Abstract

This paper presents an automated methodology able to register laser scanning point clouds using their panoramic images derived from intensity values or RGB data, the latter obtained from a co-registered camera. Starting from the panorama of each laser scan, a Feature-Based Matching (FBM) algorithm is pairwise applied to extract corresponding key-points. Robust estimators are then used to remove outliers through a generalized rejection procedure encompassing several geometric models. After tracking the twofold key- points across different scan pairs in order to increase the local redundancies, a global Least Squares block adjustment is computed for all scans. Ground control points can also be included at this stage for datum definition and control of block’s stability. The proposed method was tested on real case studies and the experiments showed that the procedure is able to deliver the registration of all scans in a fully automatic way. On the other hand, if a higher accuracy is required this solution needs a further ICP refinement.