Object-based video processing using forward-tracking 2-d mesh layers

Abstract

The invented method involves the object-based processing of parts of video frames referred to as Video Object Planes using 2-D meshes, wherein the color and shape information associated with the Video Object Planes are assumed to be known at every frame and wherein each video object is processed independently. The invented method more particularly involves utilization of the Alpha Planes, which contain the shape information, in object-based design of an initial 2-D mesh, wherein an Alpha Plane is used to form a constraining polygonal mesh boundary, as well as in object-based tracking of mesh node points, wherein motion vectors of nodes on the mesh boundary are constrained so that these node points always lie along the Alpha Plane boundary, by means of restriction of the search space or back-projection, and mesh-based Video Object Plane mapping takes into account any differences between the mesh boundary and the Video Object Plane boundary. Such invented methods may be computer-implemented or computer-assisted, as by being coded as software within any coding system as memory-based instructions executed by a microprocessor, PC or mainframe computer, or may be implemented in hardware such as a state machine.