Warped-Skip Mode For 3d Video Coding

Upcoming display technologies like auto stereoscopic displays require synthesis of virtual viewpoints to allow for depth impression without requiring the viewer to wear glasses. View synthesis relies on the availability of depth information at the receiver side to warp the recorded video data to arbitrary viewpoints. This depth data contains information about the 3D position of recorded pixel values and therefore inherently also contains the pixel-wise disparity information between multiple views to be coded. In this paper we propose to use this depth information also for compression of multi view video+depth sequences. By warping the base view's frame of a multi view video to the position of the enhancement views' camera a prediction signal is formed, which can either be used for the proposed warped skip mode or as a prediction signal to reduce the residual's energy.