Scene-Aware Video Modeling and Compression

We describe a video compression methodology that exploits the structure of the data formation process, whereby the "source\´\´ is the scene, and the ``channel\´\´ includes scaling and occlusion phenomena that are critical elements of image formation. Thus our scheme involves occlusion detection, optical flow computation, texture/structure partition, and a notion of proper sampling. We show preliminary but promising results that exceed baseline compression performance, albeit at an increased computational cost.