Registration of Point Cloud Data Based on PSO

At present, 3-D model can be build through point clouds data which is quickly get by laser scanning technology. Most of the researches on registration problems are focused on feature matching method. This paper discusses a method with global optimization algorithm. In order to solve the problem efficiently, we present a method using PSO algorithm with Cross Products of Normal Vector. This method does not require artificial targets and feature points frequently. The fitness function is used the sum of normal vectors' cross products to define, and it applies an efficient "Universal Search" and implements cloud data registration based on the best registration criterion. The application and analysis on example indicates the proposed method is effective and feasible.