Unsupervised video object segmentation and tracking based on new edge features

Abstract

We present an efficient video segmentation and tracking strategy based on edge information to assist object-based video coding, motion estimation, and motion compensation for MPEG-4 and MPEG-7. The proposed algorithm utilizes the human visual perception to provide edge information. Three parameters are introduced and described based on edge information from the analysis of a local histogram. An edge function is defined to generate the edge information map, which can be thought as the gradient image. Then, an improved marker-based region growing and merging techniques are derived to separate the image regions. An efficient temporal segmentation and tracking algorithm is also developed in time domain when the initial segmentation is given. The proposed algorithm is tested on several standard sequences and demonstrates high reliability for video object segmentation and tracking.