Abstract: In order to obtain the ideal hole repairing result of point cloud data, aiming at the defects existing in the current algorithms, we proposed a hole repairing algorithm for point cloud data based on least square support vector machine. First, the hole repairing range was estimated according to the boundary of scattered point cloud, and then according to information of hole and surrounding points, we built a surface by least square support vector machine, and repaired the hole in the point cloud data. Finally, the simulation experiment was realized by C++ language programming. The experimental results show that the least square support vector machine can effectively repair various complex holes, and the repair effect is better than other algorithms.

Key words: hole repairing, threedimensional imaging, least square support vector machine (LSSVM), surface reconstruction, point cloud data