Precise Registration of Point Clouds for Projected Fringe Image

An object has to be measured to recover its 3D shape in reverse engineering applications. The object surface is sampled point by point using a fringe projection to obtain point clouds. Inheriting and optimizing algorithm is proposed to complete precise registration of point clouds. A mathematical model is established to obtain the required transformation parameters. The realizing procedure of the proposed algorithm is described and analyzed. 6 groups of point clouds of a car flashing are acquired separately by a 3D scanning instrument using projected fringe, and point clouds registration based on the proposed algorithm is implemented. Registration error is analyzed by a multi resolution hierarchy method. It shows that the proposed algorithm can precisely register unorganized point clouds. The required number of iterations is reduced by more than 5 times compared with that of an iterative closest point algorithm. The average error is only 0.62 mm, and the registration of 2 point clouds can be completed in 2 s.