Region-Based Adaptive Distributed Video Coding

The practical application of DVC is referred to as
Wyner-Ziv video coding (WZ) where the side information
available at the decoder is an estimate of the original frame. The compression is achieved by sending the information that is needed to correct this estimation. Due to the limitation of side information prediction, the predicted frame is expected to have various degree of success along the predicted frame. In this work, we propose partitioning the considered frame into many coding units (region) where each unit is encoded differently. This partitioning is, however, done at the decoder while
generating the side-information and the region map is sent over to encoder at very little rate penalty. The partitioning allows allocation of appropriate DVC coding parameters (virtual channel, rate, and quantizer) to each region. The proposed solution preserves the simplicity of the video encoder by performing the frame partitioning process at the decoder during side information generation process. Experimental results show that adaptive DVC parameters selection per region enhances the overall PSNR when compared with frame based.