”Mixed-State Markov Random Fields for Motion Texture Modeling and Segmentation”

TOMAS CRIVELLI; BRUNO CERNUSCHI-FRÍAS; PATRICK BOUTHEMY; JIAN-FENG YAO

Atlanta, Georgia, USA

Congreso; IEEE ICIP'06, “IEEE International Conference on Image Processing”, Atlanta, Georgia, USA, 8 al 11 de Octubre 2006. En las actas del congreso en pp. 1857-1860.; 2006

IEEE

The aim of this work is to model the apparent motion in image sequences depicting natural    dynamic scenes. We adopt the mixed-state Markov Random Fields (MRF) models recently   introduced to represent so-called motion textures. The approach consists in describing  the spatial distribution of some motion measurements which exhibit mixed-state nature:  a discrete component related to the absence of motion and a continuous part for   measurements different from zero. We propose several significative extensions   of this model. We define an original motion texture segmentation method   which does not assume conditional independence of the observations for   each texture and normalizing factors are properly handled. Results on real   examples demonstrate the accuracy and efficiency of our method.