Automatic Video Object Segmentation based on New Edge Features

B.-G. Kim and D.-J. Park (Korea)


Edge Features, Video Object, Histogram Analysis, Segmentation


In the MPEG-4 and MPEG-7 video coding standards, the core technology for development is content-based function ality. Since the scheme for image data compression should be implemented in terms of independent video object rather than image frame, this paper deals with an efficient intra frame segmentation strategy based on edge information to assist object-based video coding, motion estimation and motion compensation. The suggested algorithm classifies an arbitrary frame into edge-contained blocks and non edge contained blocks at a block level. To utilise the edge information, the edge strength is defined and described by using local histogram analysis. To segment the detected edge-contained blocks with pixel accuracy, the probabilis tic background extraction algorithm is devised by using a background model which is based on the first order Markov field. The provided algorithm has a low computational cost, since the edge information can be acquired on the basis of the local histogram analysis only. The proposed technique is tested on several standard sequences and demonstrates the efficiency of video object segmentation.

Important Links:

Go Back