Interscale Predictive Wavelet Coding with Huber Markov Random Field

An interscale predictive wavelet coding (IPWC) scheme is proposed in this paper. The image is encoded top-down, from the coarsest scale to the finest scale. At each scale, the current scale coefficients are predicted from the coded coefficients at the coarser scale and the prediction residue is encoded. The interscale prediction is based on MAP estimation with Huber Markov random field statistical image model. The efficiency of IPWC is supported by experimental results.