Generalized Coherent Point Drift With Multi-Variate Gaussian Distribution and Watson Distribution

This letter introduces a novel rigid point set registration (PSR) approach that accurately aligns the pre-operative space and intra-operative together in scenario of computer-assisted orthopedic surgery (CAOS). Motivated by considering anisotropic positional localization noise utilizing undirected normal vectors sets (PSs), multi-variate Gaussian distribution Watson are utilized to model vectors’ error distributions respectively. In proposed approach, with above probability distributions, PSR problem is then formulated as maximum likelihood estimation (MLE) solved under expectation maximization (EM) framework. Our contributions three folds. First, aligning generalized points formally probabilistic manner. Second, MLE EM Third, gradients associated objective functions respect desired parameters computed provided. Experimental results on both human pelvis femur models demonstrate potential clinical values owns significantly improved performances compared existing methods.