α-channel compression in video coding

Traditional video coding research has been focused on the compression of luminance and chroma components of video signals. Lately, object-based coding algorithms also address the compression of the shape of video objects. Shape data are often represented by binary masks. In this paper, we introduce yet another dimension to video coding, namely, the α-channel compression. The α-channel has been widely used in the computer graphics world for a long time and is indispensable to digital video editing and processing. In some sense, the α-channel can be considered as a generalization of binary shape. The compression of α-channel emerges as an interesting and challenging problem in video coding research. In this paper, we begin with presenting the characteristics of this problem as we have observed in our work. We then described a shape and texture decomposition approach to this problem and study the rate-distortion property of α-channel coding