Break and Splice: A Statistical Method for Non‐Rigid Point Cloud Registration

3D object matching and registration on point clouds are widely used in computer vision. However, most existing cloud methods have limitations handling non-rigid sets or topology changes (e.g. connections separations). As a result, critical characteristics such as large inter-frame motions of the may not be accurately captured. This paper proposes statistical algorithm for registration, addressing challenge without need to estimate correspondence. The uses novel Break Splice framework treat challenges reproduction process Dirichlet Process Gaussian Mixture Model (DPGMM) cluster pair sets. Labels assigned source set with an iterative classification procedure, is registered target same labels using Bayesian Coherent Point Drift (BCPD) method. results demonstrate that proposed approach achieves lower errors efficiently registers undergoing motions. evaluated several data various qualitative quantitative metrics. outperforms state-of-the-art methods, achieving average error reduction about 60% time 57.8%.