Moving Object Segmentation for Non-rigid-object by Bayesian Clustering

摘要

An algorithm for the segmentation of the non-rigid-object with accurate boundary using spatio-temporal information is proposed in this paper. The algorithm is composed of three parts: temporal segmentation, spatial segmentation, and region merging similar to a number of existing techniques. However, background register and an improved K-Means algorithm are used in the first and second part separately. After that, Bayesian clustering is used to merging the region. The experimental results of the proposed algorithm show the algorithm is efficient and accurate.