Systems for disparity-based multiple-view interpolation

Viewpoint adaptation from multiple-viewpoint video captures is an important tool for telepresence illusion in stereoscopic presentation of natural scenes, and for the integration of real-world video objects into virtual 3D worlds. This paper describes different low-complexity approaches for generation of virtual-viewpoint camera signals, which are based on disparity-processing techniques and can hence be implemented with much lower complexity than full 3D analysis of natural objects or scenes. A realtime hardware system, which is based on one of our algorithms, has already been developed.