Scalable video coding using Symmetric Residue Pyramid and rate-distortion optimization

摘要

Videos are often required to be transmitted over different media. Scalability is an important aspect of video transmission. In case of challenged network, graceful reconstruction from a truncated bitstream becomes an issue. In order to leverage capabilities of standard video codecs, we suggest our scalable coding through a wrapper on top of any Commercial Off-the-Shelf (COTS) codecs. A single video is encoded as a set of smaller subvideos. The wrapper performs splitting and composingof subvideos. A Rate-Distortion (RD) analysis has been done to allocate optimal rates to individual subvideos. This optimal allocation minimizes overall distortion. In this paper, a hybrid scalable format which uses Symmetric Residue Pyramid for reconstruction is proposed. Superior PSNR values and visual quality have been achieved in comparison to the techniques involving other up-sampling methods.