Aiming at the problems of many existing point cloud registration algorithms, such as different registration speed and registration accuracy, this paper proposes a registration algorithm combining DNSS with Point-to-Plane ICP. It uses DNSS to extract the key points of the source point cloud data, uses the key point constraints to find the corresponding matching point pairs, and calculates the optimal rigid body transformation matrix combined with the point to plane error measurement method, so as to complete the point Cloud registration. The experimental results show that the registration accuracy of point cloud registration algorithm based on DNSSand Point-to-Plane ICP is higher than that of point-to-point ICP algorithm and Point-to-Plane ICP algorithm, and this method has significant advantages in processing point cloud data with complex geometric features and obvious features.