Signal compression, digital watermarking and pattern recognition are examples of applications of vector quantization (VQ). A relevant problem concerning VQ is codebook design. In this paper, an alternative is presented for accelerating the fuzzy K-it Means algorithm applied to codebook design. Simulation results involving VQ of images and Gauss-Markov signals show that the proposed method leads to an increase of convergence speed (reduction of the number of iterations) of the fuzzy K-it Means algorithm without sacrificing the quality of the designed codebooks.
Vector quantization; fuzzy K-Means; image coding