Measurement Axis Searching Model for Terrestrial Laser Scans Registration

Nowadays, terrestrial Lidar scans can cover rather a large area; the point densities are strongly varied because of the line-of-sight measurement principle in potential overlaps with scans taken from different viewpoints. Most of the traditional methods focus on registration algorithm and ignore searching model. Sometimes the traditional methods are directly used to align two point clouds; a large critically unsolved problem of the large biases will be created in areas distant from the overlaps while the local overlaps are often aligned well. So a novel measurement axis searching model (MASM) has been proposed in this paper. The method includes four steps: the principal axis fitting, the measurement axis generation, low-high-precision search, and result generation. The principal axis gives an orientation to the point cloud; the search scope is limited by the measurement axis. The point cloud orientation can be adjusted gradually until the achievement of the global optimum using low- and high-precision search. We perform some experiments with simulated point clouds and real terrestrial laser scans. The results of simulated point clouds have shown the processing steps of our method, and the results of real terrestrial laser scans have shown the sensitivity of the approach with respect to the indoor and outdoor scenes.