A hybrid graphical model for robust feature extraction from video | IEEE Conference Publication | IEEE Xplore