Optimal estimation for error concealment in scalable video coding

This work is concerned with scalable video coding and, specifically, with error concealment for the mitigation of damage caused by packet loss at the enhancement layer. We propose a frequency-domain error concealment scheme that makes full use of the information available at the decoder from both the current base layer, and prior enhancement layer frames. It employs a statistical model for the evolution of transform coefficients from frame to frame and implements the optimal estimate of the reconstructed coefficient. Moreover, the ideas underlying the proposed approach allow detection and correction of blocks that suffer from significant error propagation from past losses. Substantial PSNR gains are observed in the experiments.