3DMNDT: 3D Multi-View Registration Method Based on the Normal Distributions Transform

The normal distributions transform (NDT) is an effective paradigm for point set registration. This method was initially designed pair-wise registration and suffers from the accumulated error problem when directly applied to multi-view Under framework of point-to-cluster correspondence, this paper proposes a novel named 3D based on (3DMNDT), which integrates k-means clustering Lie algebra optimizer achieve More specifically, cast into maximum likelihood estimation problem. Firstly, utilized divide all data points different clusters, where one distribution computed locally model probability measuring in each cluster. Subsequently, formulated by NDT-based function. To maximize function, introduced developed optimize rigid transformation sequentially. 3DMNDT implements clustering, NDT computing, optimization alternately until desired results are obtained. Experimental tested benchmark sets illustrate that can state-of-the-art performance Note Practitioners —This motivated solving registering multiple sets. well-known widely robotic domain. extends original simultaneously align more than two integrated estimate parameters. demonstrate its superior accuracy, efficiency, robustness