Study on the Plane and Spherical Feature Extraction Based on Differential Information of Point Clouds

Abstract

Geometric features play a very important role in the mechanical product reverse modeling. Plane and spherical features are widely used in the product design. A method for determining plane and spherical features is presented in the paper. The difierential information of point clouds, such as normal vectors and curvatures, is calculated. Least square fltting algorithm is used to calculate unit normal vectors of point clouds. Quadratic parabolic surface model is used to calculate curvatures of point clouds. Unit normal vectors are used to establish the Gauss map, and curvatures are used to establish the normal curvature map. Because the normal vector of the plane is a constant, two normal curvatures of the sphere are equal in absolute values, the plane and spherical features of point clouds are able to quickly be extracted. Experimental results showed that the proposed method is efiective.