Aiming at the problem of taking a lot of time to iterate, easy to converge to the local optimal, and long time for the point cloud registration, proposes a point cloud registration method based on the optimization of the corresponding relationship of the paired point clouds. Firstly, using the fast feature histogram (FPFH) and feature optimization to generate the initial correspondence; then, by checking whether the corresponding points meet the nearest neighbor principle, the initial corresponding point set is obtained, and then the initial point set is judged by the L2 norm ratio, and the corresponding point set with the correct corresponding relationship is basically selected. Finally, the point cloud is calculated by alternating optimization Transformation matrix to achieve precise registration of point clouds. Experimental results show that compared with the traditional ICP algorithm，the proposed algorithm has fast registration speed, high registration accuracy, and it is robust to point clouds with low overlap and high noise.