Abstract

We propose a depth-assisted error concealment method for slice losses in intra frames of 2-D+depth encoded 3-D video. Intra frames in the depth sequence are offset from intra frames in the 2-D view sequence, such that the corresponding frames in the other sequence are not also intra mode. For slice losses in an intra frame of the 2-D sequence, motion vector (MV) candidates to conceal the lost macroblocks come from the co-located MVs in the depth frame as well as from MVs in the previous 2-D frame chosen based on motion information from the depth frame. Then the temporal smoothness of the depth sequence is used to filter out MV candidates. Finally, an enhanced distortion criterion is used for MV selection, based on contour information in the depth frame and boundary matching. Experimental results show that the proposed method provides improved performance over existing methods.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call