Registration for multi-station point cloud of terrestrial laser scanning

Pair-wise registration is a common method for registration of multi-views point clouds. Owing to the characteristics of error propagation and accumulation, using pair-wise registration method will result in severe distortion for object researched. In order to overcome this shortcoming, we present a new registration method. Firstly, we adopted coordinate system of first scan as a uniform coordinate system, used pair-wise registration method to calculate transformation parameters of adjacent two scans, and connected all point clouds into a loose network. Secondly, based on the common points that point cloud is adjacent to, and conditions that the first scan must have the same coordinates with real scanning coordinates in the first scan coordinate system, and common points between adjacent point clouds should haves the same coordinates in uniform system, we used nonlinear correction model to compute distortion parameters of each point cloud. And lastly we corrected distortion of each single point clouds and determined their best positions. An experiment was carried out and the result showed the correctness of the method.