一种基于覆盖域密度的LBG算法

Linde-buzo-gray algorithm based on coverage density

针对矢量量化中Linde-Buzo-Gray(LBG)算法产生大量无效或重复码向量问题,提出了一种基于覆盖域密度分割码向量方法。在LBG算法的更新迭代步骤,记录各码向量的覆盖域及其平均失真率,计算覆盖域密度;码向量分割时,只对"坏的"码向量进行分割,"好的"码向量直接复制到下一代码书中。实验表明,修改后的算法提高了LBG的鲁棒性,码书质量也得到一定程度提高。

To alleviate the problem that codebook generated by Linde-Buzo-Gray(LBG) algorithm includes a lot of invalid or identical codevectors in vector quantization,a coverage density based LBG algorithm is proposed.Coverage distortion and density for each codevectors are calculated in the iteration step of LBG.In splitting step,only bad codevectors with lower coverage density are split into two and invalid codevectors are deleted.Good codevectors,which with very higher coverage density,are insert into the new codebook without any change.Experiment results show that the modification to LBG improve it's robustness and the quality of codebook.