Projective distribution entropy and point clouds mosaic algorithm

The multi-view point clouds mosaic IS an important technology in ladar data preprocessing, and it is also an open problem in 3D information processing. In the paper, a projective distribution entropy based point clouds mosaic algorithm is proposed. First, a unique coordinate system is used to estimate the space transformation between the multi-view point clouds, and to perform the coarse mosaic. Second, the Iterative Closest Point (ICP) method is applied for the fine mosaic. In order to enhance the robustness of the ICP method, the deterministic annealing algorithm is used. Experiments on complex point clouds demonstrated that the algorithm was reliable and effective.