Universal lattice-based quantizers for multiple descriptions

Multiple description source coding is aimed at achieving graceful degradation in reconstruction with respect to losing portions of the code, with the cost of some redundancy. In our research we examine source coding schemes for two descriptions and three decoders based on entropy coded dithered ("universal") quantizers (ECDQ). We propose two techniques. The first is a two stage encoder, where the first stage produces two coarse descriptions, and the second stage produces a refinement code to be used only by the joint decoder. The second technique is a dithered and periodic version of a scalar quantizer with distributed cells for multiple descriptions proposed by Vaishampayan (1993). Both techniques ae shown to be equivalent to additive noise channels. Analytical expressions for their performance are derived and compared to the optimal rate regions at both high and low resolutions. The first technique is conceptually simple and easily tuned to various parameters, but is inherently sub-optimal. The second technique is less versatile but shows promising results that resemble those of the ECDQ for single description encoding.