05673884

j

05673884 Xiao, Chuanmin Shi, Ze-lin Qi, Lin Moving object segmentation technology based on DA-STMRF model. J. Comput. Appl. 28, No. 9, 2440-2442 (2008). 2008 Science Press, Beijing ZH image segmentation Markov random field (MRF) discontinuity-adaptive

doi:10.3724/SP.J.1087.2008.02440

http://www.computerapplications.com.cn/EN/abstract/abstract11854.shtml

Summary: In order to avoid the drawback of over-smoothness in the conventional Spatiotemporal Markov Random Field (MRF) model, a moving object segmentation technology based on Discontinuity Adaptive-Spatiotemporal Markov Random Field (DA-STMRF) model is proposed. Initial labels are derived after frame difference images are converted to binary images, and the AND-label is obtained with the AND-operation on the two initial labels. After constructing corresponding energy functions, we used metroplis sampler algorithm to optimize the label field. Segmentation experiments on several image sequences show that the algorithm proposed is effective.