A wavelet-based multiple-description coding combining pairwise correlating transform with quincunx sub-sampling

Abstract

This paper proposes a novel and efficient two-description image coding scheme in the wavelet domain. Here, each description includes one basic description produced by pairwise correlating transform (PCT) and one enhanced description formed by quincunx sub-sampling. First, we partition the low-frequency wavelet coefficients into two groups. Then the PCT was applied between the two groups to generate two new correlated groups as two basic descriptions, which are coded and transmitted through two separate channels. Second, we can get two enhanced descriptions from the high-frequency wavelet coefficients of the highest-level decomposition by quincunx sub-sampling. And the Wavelet Domain Interpolation for tree Reconstruction (WDIR) is used to conceal coefficients loss in Wavelet-Trees when one enhanced description lost during transmission. Experimental results demonstrate that the proposed MDC scheme can achieve good coding performance.