Frame Loss Concealment for Stereoscopic Video Plus Depth Sequences

Abstract

In this work, an efficient frame loss concealment algorithm for stereoscopic video plus depth sequences is proposed. To conceal an erroneous left color frame, we extrapolate motion vectors for the temporal error concealment. To conceal a left depth frame, we exploit the motion vectors of the corresponding left color frame. To conceal a right color frame, we employ a 3D image warping technique to determine matching pixels between inter-views, and perform the reconstruction based on the similarities of the motion vectors and the intensity differences of matching pixels. Finally, to conceal a right depth frame, we use the motion vectors of the right color frame and the left color frame. Simulation results show that the proposed algorithm is more effective than conventional algorithms in recovering erroneous color and depth frames. Moreover, it is also demonstrated that the proposed algorithm can be used for efficient virtual view synthesis in error prone environments.