Registration of LiDAR Data and Dense-Matching Point Cloud in Forest Areas

Abstract

The use of georeferenced data without control points leads to registration failures of sequent aerial images and airborne LiDAR data of forest areas. To address this problem, this work proposes a registration method of sequent images and LiDAR data captured by airborne sensors in forest areas. Firstly, 3D points are generated from sequent images by a multi-view stereo algorithm. Secondly, bare-Earth DEMs are created by a filtering method applied to the cloud points and contours are used to initially register the DEMs. Finally, dense-matching data are triangulated into a TIN and interpolated at each horizontal coordinate of the LiDAR data. The transformation matrices between the two cloud points are calculated using LiDAR data and interpolated points. By using this method, I register the two kind of DSMs successfully.