Point cloud registration algorithm based on NPFC neighborhood feature descriptor

Point cloud registration technology is widely used in machine vision, reverse engineering, and other neighborhood applications, and its efficiency and accuracy have an important impact on product data models. Aiming at the problems of the traditional ICP registration algorithm, such as slow convergence speed and high requirements on the initial point cloud position, this paper proposes a point cloud registration algorithm based on the neighborhood point feature covariance matrix descriptors. First, the key points of the point cloud data are extracted by combining the average angle of the neighborhood normal vector and ISS algorithms; then, the NPFC descriptors of the key points are calculated, and the two-way nearest neighbor feature matching is performed according to the similarity of the NPFC to obtain the initial correspondence; for the initial correspondence, the RANSAC algorithm is used to reject the mismatches to obtain the final correspondence pairs, and the initial registration parameters are calculated by using the final correspondence; finally, the iterative nearest point algorithm is used to perform the fine registration. Experiments on public datasets show that NPFC descriptor has high descriptiveness and robustness in the face of noise. The registration results also confirm the superiority of our registration method in terms of accuracy and efficiency.