LBG与SOFM应用于矢量量化的比较研究

Comparative Study on the Application of LBG and SOFM in Image Vector Quantization

在介绍矢量量化以及LBG算法和SOFM算法的基础上,通过实验对比了LBG算法和SOFM算法在应用于图象矢量量化压缩过程时,码书大小、码字大小以及初始码书生成方式等因素对图像压缩性能的影响,得到了相关结论:固定码字矢量维数,码书越大,压缩比越小,重建图像质量越好;固定码书,码字矢量维数越小,编码性能越好;LBG算法对初始码书敏感,而SOFM算法由于所具备的自适应特性对初始码书不敏感。论文最后提供了一些改进思路,为改进传统矢量量化算法及设计新的矢量量化算法以提供了参考。

Based on the introduction of vector quantization（VQ）,LBG and SOFM algorithm,this paper compared the performance of these two algorithms of codebook design in VQ by using the indexes of peak signal-to-noise ratio,compression ratio and coding time,all of which are influenced by factors such as the size of codebook and codeword,and initial codebook.The results are as follows： by a given size of codeword,a larger codebook will result in a better compression ratio and image reconstruction quality;Also,by the given size codebook,a smaller codeword will result in a better coding performance;LBG algorithm is sensitive to initial codebook while the SOFM algorithm isn＇t because of its self adaptive characteristics.At the end of the paper,thoughts of improvement are given,which provide a reference for designing new vector quantization algorithm and advancing traditional vector quantization algorithm.