Multiple description coding by successive quantization

We propose two structured multiple description (MD) vector quantization schemes with an iterative technique to design the codebooks and partitions. The schemes are derived from the recent theoretical work by Chen et al. [6]. In the first scheme, the central decoder is formed by weighted sum of the side codebooks whereas the second scheme employs the optimum central decoder. The objective of the proposed iterative method is to minimize a Lagrangian cost function (defined as the weighted sum of the central and side distortions) to jointly design the side codebooks and find the associated partitions. The optimal parameters of the problem are also found to minimize the central distortion. We demonstrate by simulations that our proposed methods perform very closely to the unstructured, full search MD quantizer with considerably less complexity and with a few iterations.