A Novel Temporal Error Concealment Base on Statistical Pixel Estimation Model

The traditional temporal error concealment methods are recovery-of-lost-MV based. However, in this paper a statistical pixel estimation model is established, based on which a novel temporal error concealment method is proposed. The proposed method has two advantages over the traditional methods as follows. Firstly, the correlation of the motion vector (MV) field is fully exploited to recover the lost block and no MV needs to be explicitly estimated. Secondly, less error energy of the previous frame is propagated to the successive frames when compared to the traditional methods. Simulation results show that 1-3 db or so improvement on the average in PSNR can be obtained when error propagations are considered