Title :
Dynamic background discrimination with belief propagation
Author_Institution :
Res. Inst. of Comput. Sci. & Technol., Ningbo Univ., China
Abstract :
A probabilistic graphical model is proposed for the complex video foreground and background discrimination. The model learns the temporal and the spatial correlation from the video input data. The inference of the graphical model is achieved with the generalized belief propagation algorithm. Experiments have shown that the proposed method is able to model the dynamic backgrounds containing swaying trees, bushes and moving ocean waves. The final segmentation results are very promising.
Keywords :
belief maintenance; correlation theory; image motion analysis; image segmentation; image sequences; inference mechanisms; learning (artificial intelligence); probability; spatiotemporal phenomena; video signal processing; bushes; complex video foreground discrimination; dynamic background discrimination; generalized belief propagation; graphical inference model; image segmentation; image sequence; learning mechanism; ocean waves; probabilistic graphical model; spatial correlation model; swaying trees; temporal correlation model; video input data; Bayesian methods; Belief propagation; Computer vision; Graphical models; Hidden Markov models; Image processing; Image segmentation; Markov random fields; Random variables; Uncertainty;
Conference_Titel :
Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference on
Print_ISBN :
0-7803-8403-2
DOI :
10.1109/ICMLC.2004.1384600
