图像压缩的小波系数模型

Embedded Image Coding on Wavelet Coefficients

通过对 3种经典的小波系数模型 (EZW模型 ,SPIHT模型 ,EBCOT模型 )进行仿真实验 ,表明SPIHT的压缩性能优于EZW(平均高 0 .6dB) ,而EBCOT的压缩性能又优于SPIHT(平均高 0 .5dB) ,再通过对 3种模型的分析 ,明确了 3种模型性能差异的根源在于对子带内小波系数的相关性和子带间小波系数的相似性的不同利用。最后 ,综合上述 3种模型的优点 ,充分利用子带内小波系数的相关性和子带间小波系数的相似性 ,提出了一种新型小波系数模型。实验表明该模型在压缩比方面优于上述 3种模型。

The modeling of wavelet coefficient for still image compression is studied.Firstly,through simulation experiments on three classical wavelet coefficient models(i.e.,EZW model,SPIHT model/and EBCOT model),it is shown that SPIHT is superior to EZW in terms of compression performance(0.6 dB higher)and the compression performance of EBCOT is significantly better than that of SPIHT on average. Then the cause for such performance differences is explained through the analysis of models.Finally a new wavelet coefficients model is proposed which combines the advantages of the three classical models.Experimental results show that our model is better than the above mentioned three models with respect to compression ratio.