Design of low-resolution multiple description vector quantizers by means of the Self Organizing Maps

Multiple description coding is an appealing tool to guarantee graceful signal degradation in the presence of unreliable channels. While the principles of multiple description scalar quantization are well-understood and solid guidelines exist for the design of effective systems, the same does not apply to vector quantization, with burdensome and not always satisfactory design techniques. In this work we show that good multiple description VQ codebooks can be designed in a simple and efficient way by resorting to the Self-Organizing Maps algorithm, where just a few parameters must be reasonably selected. Experimental results show that the proposed technique has a performance comparable to a well-known reference technique based on greedy optimization, but a much lower computational burden. In addition, the resulting codebook can be itself optimized, thus providing even better performance.