Image Segmentation with Pseudo-marginal MCMC Sampling and Nonparametric Shape Priors

TL;DR

This paper introduces an efficient pseudo-marginal MCMC method for image segmentation that leverages nonparametric shape priors, enabling scalable sampling from complex posterior distributions and capturing multiple segmentation modes.

Contribution

It presents a novel pseudo-marginal MCMC approach that scales independently of training set size and effectively characterizes multimodal shape posteriors in image segmentation.

Findings

Scales well with large datasets

Capable of sampling from multimodal distributions

Demonstrates effectiveness on experimental data

Abstract

In this paper, we propose an efficient pseudo-marginal Markov chain Monte Carlo (MCMC) sampling approach to draw samples from posterior shape distributions for image segmentation. The computation time of the proposed approach is independent from the size of the training set used to learn the shape prior distribution nonparametrically. Therefore, it scales well for very large data sets. Our approach is able to characterize the posterior probability density in the space of shapes through its samples, and to return multiple solutions, potentially from different modes of a multimodal probability density, which would be encountered, e.g., in segmenting objects from multiple shape classes. Experimental results demonstrate the potential of the proposed approach.