Title :
Fusion of geometric models from VLS overlapping profiles
Author :
Lin, Yi ; Hyyppä, Juha
Author_Institution :
Dept. of Remote Sensing & Photogrammetry, Finnish Geodetic Inst., Masala, Finland
Abstract :
This study advances a new method for fusion of geometric models from the overlapping profiles collected by vehicle-based laser scanning (VLS), which is developed as a novel mapping technique. The schematic starts from line segment extraction based on random sample consensus (RANSAC) for each profile, and the resulted line segments are grouped. Then, the grouped line segments are fitted to local planes, and the local planes involved in different profile indicators are fused based on maximum a posteriori (MAP) estimation. Finally, the achieved 3D models are refined under the trial-and-error strategy. The mean distances from the fused local planes to the related inlier points are all less than 0.3 m, and numerical analysis based on the real-measured VLS data has validated the new method.
Keywords :
feature extraction; laser ranging; maximum likelihood estimation; sensor fusion; solid modelling; geometric models; line segment extraction; local planes; maximum a posteriori estimation; overlapping profiles; profile indicators; random sample consensus; trial-and-error strategy; vehicle-based laser scanning; Clouds; Estimation; Fitting; Laser modes; Object oriented modeling; Solid modeling; Three dimensional displays; Fusion; RANSAC; VLS; geometric model; overlapping profiles;
Conference_Titel :
Geoscience and Remote Sensing Symposium (IGARSS), 2010 IEEE International
Conference_Location :
Honolulu, HI
Print_ISBN :
978-1-4244-9565-8
Electronic_ISBN :
2153-6996
DOI :
10.1109/IGARSS.2010.5650072