A multi-scale covariance matrix descriptor and an accurate transformation estimation for robust point cloud registration

Abstract

This paper presents a robust point cloud registration method based on a multi-scale covariance matrix descriptor and an accurate transformation estimation. Comparing with state-of-the-art feature descriptor such as FPH, 3DSC, Spin Image, etc, our proposed the multi-scale covariance matrix descriptor is superior to deal with registration problem under higher noise environment, which is since mean operation in generating covariance matrix can filters out most of the noise-damaged samples or outliers and also makes itself be robust to noise. Comparing with transformation estimation such as feature matching, clustering, ICP, RANSAC, etc, our transformation estimation is able to find a better optimal transformation between a pair of point clouds which is since our transformation estimation is a multi-level point cloud transformation estimator including feature matching, coarse transformation estimation based on clustering and a fine transformation estimation based on ICP. Experiment findings reveal that our proposed feature descriptor and transformation estimation outperforms state-of-the-art feature descriptors and transformation estimation, and registration effectiveness based on our registration framework of point cloud is extremely successful in Stanford 3D Scanning Repository, SpaceTime dataset, and especially Kinect dataset.