Skip to Main Content
This paper proposes a novel method that utilizes inter-view correlation in order to reduce the bitrate required to represent the prediction mode including the intra prediction direction, the motion vectors, the reference picture indexes, and the sub-block partitioning. The proposed method derives these kinds of macroblock (MB) information at the decoder side by comparing each image signal predictor with the view synthesis pictures. To accurately evaluate the predictors, we also propose a measure that minimizes the impact of errors in the depth information and view-dependent image signals. Experiments show that the proposed method reduces the bitrate by up to 25% relative to the international multiview video coding standard, and about 20% relative to the conventional view synthesis prediction method.