Super-resolution for compressed video based on DCT side information

In previous papers, frequency-domain quantization noise was always treated as uniform distribution in super-resolution construction for compressed video, when at high compressed ratio, this model may introduce error. This paper provides a Laplace model of DCT coefficients whose parameters can extract from the encoder, then transfers them as side information to the decoder, and utilizes the distribution of DCT coefficients as priori information in MAP (Maximum a Posterior) for super-resolution construction. Experimental results demonstrate the effectiveness of this approach.