An improved ICP registration algorithm for 2D point cloud

In this study, an improved iterative closest point (ICP) registration algorithm was proposed to address such problems as great time consumption of iterative calculations and relatively large registration errors in the traditional ICP algorithm within the field of 2D point cloud registration. During the iteration process, the point-to-line (PL) objective equation was simplified to solve the linear least squares fast through singular value decomposition, etc. Subsequently, this result could be further nonlinearly optimized, and the weighted objective functions for the PL distance and point-to-point (PP) distance were introduced in the optimization process. Moreover, a simulation experiment was performed using point clouds that were randomly generated multiple times. The results reveal that the improved ICP algorithm proposed in this study can improve both the registration accuracy and efficiency.