Generation method of normal vector from disordered point cloud

Abstract

The normal vector generation from point cloud plays a vital role in 3D laser scanning data processing. An effective generation method of normal vector from disordered point cloud is proposed in the paper. In the proposed method, disordered point cloud is firstly projected on the 2D plane via cylindrical surface. After establishing the triangle grids, the unit normal vector of each vertex is then calculated by searching the conjoint triangles that include this vertex. The normal calculation and direction determination of each conjoint triangle are relied on the connection relation among triangular grids and the corresponding mapping relation between the 2D and 3D coordinates of the same point. Software called Seagull2.0 is compiled to realize the generation of normal vector from disordered point cloud according to the proposed method. Two experiments are carried out on Confucius statue point cloud and industry model point cloud respectively. The results indicate that the determination of normal direction is correct and the generation method is reliable and easy to realize.