A point cloud registration algorithm based on canonical correlation analysis is proposed. We centralize the target point cloud and the point cloud to be registered, and rotate it around the coordinate origin. The two sets of point clouds can satisfy the maximum square of the correlation coefficient between the dimensions. The two sets of rotation matrices are solved by typical correlation analysis method. The rotation matrix and the translation vector of the rigid transformation between the two points of the clouds are solved by the rotation matrix, and the registration of the point cloud is realized. We use the proportional square value of the eigenvalues of the covariance matrix to scale the registration point cloud proportionally, and complete the affine registration. The simulation results show that, compared with several other algorithms, the proposed algorithm can be quickly and accurately registered with good stability, when point clouds are out of order, occluded, missing, size scaling and interrupted by noise.