Abstract
The rapidity with which digital information, particularly video, is being generated by broadcast networks world over has necessitated the development of tools for compact representations of such media. In this paper, we propose a method to obtain compact visual representations for Human postures. The novelty of our technique is its ability to cope with Human postures in terms of both scalability and variability. Our algorithm uses a combination of contour analysis and skeletonization to obtain compact data for semantic descriptors of such visual information. These compact data can be used in limited bandwidth applications.
