A Robust Lin-Buzo-Gray Algorithm in Data Vector Quantization

Vector quantization (VQ) is an effective means of data compression for it maps a set of real numbers into a single integer. In this paper, a coverage density based initial codebook generating mechanism is proposed to replace the random splitting method in Linde-Buzo-Gray algorithm (LBG). Coverage distortion and density of each codevector are calculated in the iteration step of LBG. In splitting step, only bad codevectors which with lower coverage density are split into two and invalid codevectors be deleted. Good codevectors, which with very higher coverage density, are insert to the new codebook without splitting. Experiment results show that the modification to LBG improves the robustness of LBG and the average distortion is also reduced up to 7% than LBG.