Improved PSO-LBG to design VQ codebook

The purpose of this study is to improve the Linde-Buzo-Gray (LBG) algorithm by using improved particle swarm optimization (IPSO) to generate vector quantization (VQ) codebooks. The proposed IPSO-LBG algorithm improves PSO-LBG in terms of initial particles and evolution function. The IPSO generates better initial particles by grouping the original blocks and expands search area by dynamically adjusting the individual coefficient, social coefficient, and weight value of the evolution function. Then, the best solution obtained by the IPSO is provided for a LBG to find a codebook. Therefore, the proposed IPSO-LBG is well enhanced on search space and convergence. The experiment results confirm that the proposed IPSO-LBG find better codebooks than existing methods reported in the literatures.