Multiview Distributed Video Coding with encoder driven fusion

In this paper, a scheme based on multiview and Distributed Video Coding (DVC) is introduced. More specifically, a new fusion technique between temporal, homography and Disparity Compensation View Prediction (DCVP) side informations is introduced. For this purpose, a binary mask is computed at the encoder based on knowledge of the original video. This mask is compressed and then transmitted to the decoder. The latter merges the different side informations with the help of the reconstructed binary mask. The simulation results show that the fusion improves the rate-distortion performance over monoview DVC by a maximum gap of around 1.0 dB.