Point set registration with mixture framework and variational inference

Xinke Ma, Shijin Xu, Jie Zhou, Qinglu Yang, Yang Yang, Kun Yang, Sim Heng Ong

Abstract: Highlights•We define a Gaussian variational mixture model.•We develop a three-phase registration strategy.•We use anisotropic Gaussian components to weaken the effect of outliers.•We apply the Dirichlet distribution to distinguish the missing points.•The performance of our method outperforms state-of-art methods in this area.