Point Cloud Alignment Method Based on Improved ISS-ICP Algorithm

Abstract

For the traditional Iterative Closest Point (ICP) algorithm, its registration efficiency is low and the initial position of the registered point cloud is high. Accordingly, a point cloud registration method combining the optimized Intrinsic Shape Signatures (ISS) algorithm with the improved ICP is proposed. Specifically, the voxel filter is used to sample the original point cloud, then the key points are extracted by optimizing the search radius of the ISS algorithm, and described by fast point feature histogram (FPFH), and the corresponding relationship is established according to the feature. Subsequently, the normal features and the RANSAC algorithm are fused to eliminate the mismatching point pairs, and the initial transformation matrix is obtained by singular value decomposition(SVD). Finally, the ICP algorithm with median distance constraint is used to complete the precise registration. Experiments suggest that the accuracy and efficiency of the proposed algorithm are significantly improved compared with the traditional ICP algorithm.