Fuzzy channel-optimized vector quantization

A novel fuzzy clustering algorithm for the design of channel-optimized sour ce coding systems is presented in this letter. The algorithm, termed fuzzy channel-optimized vector quantizer (FCOVQ) design algorithm, optimizes the vector quantizer (VQ) design using a fuzzy clustering process in which the index crossover probabilities imposed by a noisy channel are taken into acc ount, The fuzzy clustering process effectively enhances the robustness of t he performance of VQ to channel noise without reducing the quantization acc uracy. Numerical results demonstrate that the FCOVQ algorithm outperforms e xisting VQ algorithms under noisy channel conditions for both Gauss-Markov sources and still image data.