Inexact Bayesian point pattern matching for linear transformations
作者：
Highlights：
• We model matching sets of points related by linear transformations.
• We model inexact matching (not a complete 1-1 correspondence) of noisy points.
• Variational Bayesian methods give computationally efficient posteriors for variables.
• Posterior matching probabilities indicate matched and unmatched points between sets.
• Technique is demonstrated on 3D synthetic data and optical microscopy image stacks.
摘要
Highlights•We model matching sets of points related by linear transformations.•We model inexact matching (not a complete 1-1 correspondence) of noisy points.•Variational Bayesian methods give computationally efficient posteriors for variables.•Posterior matching probabilities indicate matched and unmatched points between sets.•Technique is demonstrated on 3D synthetic data and optical microscopy image stacks.
论文关键词：Bayesian methods,Variational approximation,Point pattern matching,Iterative closest point algorithm,Linear transformation
论文评审过程：Received 16 March 2013, Revised 11 February 2014, Accepted 26 April 2014, Available online 6 May 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.04.022
