Description method for spatio-temporal regions in a video and its application

A new description method for spatio-temporal regions in a video is proposed. It is efficient in terms of description data size and computational complexity under the condition that the regions in each frame can be approximated by simple figures. It approximates the object region in a frame by a simple figure such as a rectangle, an ellipse, or a polygon, and describes the coordinates of its representative points (vertices). Then, the motions of the representative points are described by polynomial spline functions to express the regions in the following frames efficiently. The proposed method is adopted in the MPEG-7 standard to provide the functionality of spatio-temporal location description. Experimental results using synthesized and real videos are presented to show the efficiency. Also, movie-centric hypermedia, a typical application of this description method, is introduced in this article. (C) 2004 Wiley Periodicals, Inc.