A contribution to video coding using layers

Layered representations of video sequences, originally introduced in [1], are important in several applications. Specifically, they are very appropriate for object tracking, object manipulation, content-based scalability and video coding which are among the main functionalities of the future standard MPEG-4, [2] [3]. In addition they will be also useful for MPEG-7 in the corresponding image analysis part. A robust representation of moving images based on layers has been presented in [4]. The objective of this paper is to provide the first results on video coding using the segmentation scheme presented in [4]. To that end, the shape information (alpha map) has been coded using a Multigrid Chain Code approach and the layers (the texture) have been coded using a Shape Adaptive DCT scheme that is being considered as a texture coding approach in the current definition of MPEG-4. Results are provided for a variety of compression ratios and different image quality that prove the validity of the approach.