Dynamic Differential Evolution Algorithm Applied in Point Cloud Registration

Abstract

Aiming at the problem of point cloud registration, a registration method based on dynamic differential evolution algorithm (DDE) is proposed. The voxel grid method is used to uniformly sample the point cloud to reduce the time complexity of DDE. The individuals of DDE are coded by rotation angle and the translation distance. The kd-tree is used to search for corresponding point pairs, and the root mean square error is defined as the objective function of DDE. Experiments show that the proposed method can register point clouds with a large difference in position, and has better registration accuracy than iterative closest point (ICP).