基于整数平方量化阈值和索引表的小波图像编码算法

Wavelet image coding algorithm based on integer square quantization threshold and index table

一般小波变换 ,零树编码优点很多 ,但对于能量集中性较差的整型小波变换就显得力不从心 .因此对整型小波变换存在的这一特点结合零树编码的优点提出了一种新算法。第一 ,根据整型小波变换后各子带系数幅值的动态变化较小 ,小波图像能量较一般小波差的特点 ,提出了利用整数平方作为量化阈值的整数平方算法 ,充分利用了整型小波变换 ;第二 ,提出一种基于索引表和游程编码的小波零树编码的新思路 。

Zero tree coding has many advantages for wavelet transform, but it is not so good for integer wavelet transform that has a weaker energy concentration. This paper presents a new algorithm in two considerations. First, to solve the problem that integer wavelet transform has a smaller dynamic change in each sub band coefficient and integer wavelet images have a weaker energy compaction, we propose an 'integer square algorithm' which uses integer square as quantization threshold and takes full advantage of the characters of integer wavelet transform. Second, we provide a new idea of wavelet zero tree coding on the basis of index table and RLC which simplifies the encoding and decoding processes for better compressing images.