Online video segmentation by bayesian split-merge clustering

摘要: We present an online video segmentation algorithm based on a novel nonparametric Bayesian clustering method called Split-Merge Clustering (BSMC). BSMC can efficiently cluster dynamically changing data through split and merge processes at each time step, where the decision for splitting merging is made by approximate posterior distributions over partitions with Dirichlet Process (DP) priors. Moreover, sidesteps difficult problem of finding proper number clusters virtue flexibility models. naturally apply to segmentation, which composed three steps--pixel clustering, histogram-based temporal matching. demonstrate performance our complex real sequences compared other existing methods.