Empirical Bayesian EM-based Motion Segmentation

A recent trend in motion-based segmentation has been to rely on statistical procedures derived from expectationmaximization (EM) principles. EM-based approaches have various attractives for segmentation, such as proceeding by taking non-greedy soft decisions with regards to the assignment of pixels to regions, or allowing the use of sophisticated priors capable of imposing spatial coherence on the segmentation. A practical difficulty with such priors is, however, the determination of appropriate values for their parameters. In this work, we exploit the fact that the EM framework is itself suited for empirical Bayesian data analysis to develop an algorithm that finds the estimates of the prior parameters which best explain the observed data. Such an approach maintains the Bayesian appeal of incorporating prior beliefs, but requires only a qualitative description of the prior, avoiding the requirement of a quantitative specification of its parameters. This eliminates the need for trial-and-error strategies for the determination of these parameters and leads to better segmentations in less iterations.