1. INTRODUCTION
As video sources are desired to be represented by different spatial resolutions because of various channel conditions and customer device capabilities, spatially scalable video coding plays an important role in many video applications. To achieve spatial scalability, the video source lower resolution version is usually encoded independently as the base layer, and the refinement data are added as enhancement layers to provide higher resolution versions. Since the inputs of these spatial layers are highly correlated, inter-layer prediction mechanism should be applied to improve compression efficiency.