Generalized Point Set Registration With Fuzzy Correspondences Based on Variational Bayesian Inference

Point set registration (PSR) is an essential problem in surgical navigation and computer-assisted surgery (CAS). In CAS, PSR can be used to map the intraoperative space with preoperative volumetric image space. The performances of real-world scenarios are sensitive noise outliers. This article proposes a novel point approach where additional features (i.e., normal vectors) extracted from sets utilized convergence algorithm guaranteed theoretical perspective. More specifically, we formulate vectors by generalizing Bayesian coherent drift (BCPD) into 6-D scenario. proposed more accurate robust outliers, guaranteed. Our contributions this summarized as follows. 1) formally formulated through BCPD approach. 2) formulas for updating parameters during algorithm’s iterations given closed forms. 3) Extensive experiments have been done verify specifically its significant improvements over has validated.