Output details
11 - Computer Science and Informatics
University of Portsmouth
On-line learning of shape information for object segmentation and tracking
<23>Object shape tracking in video is challenging. This work proposed a novel first order Markov model to track shapes of deformable objects. The shape similarity matrices have massive potential in improving action recognition since it contains interesting characteristics of object shapes during tracking. As part of the general deformable contour frameworks this work has also led up to the projects on computer vision for traffic safety application funded by Mitsui Sumitomo Welfare Fund in 2010 (J. Chiverton, S. Uttama) and advanced medical image analysis and visualization grant (Prof. John and X. Xie) funded by NISCHR in 2011-2014.