Ming-Chieh Lee, Wei-ge Chen, Chih-lung Bruce Lin, Chuang Gu, Tomislav Markoc, Steven I. Zabinsky, and Richard Szeliski
A layered video object coding system is presented in this paper. The goal is to improve video coding efficiency by exploiting the layering of video and to support content-based functionality. These two objectives are accomplished using a sprite technique and an affine motion model on a per-object basis. Several novel algorithms have been developed for mask processing and coding, trajectory coding, sprite accretion and coding, locally affine motion compensation, error signal suppression, and image padding. Compared with conventional frame-based coding methods, better experimental results on both hybrid and natural scenes have been obtained using our coding scheme. We also demonstrate content-based functionality which can be easily achieved in our system.
In IEEE Transactions on Circuits and Systems for Video Technology