基于整数平方阈值和索引表的静止图像编码算法

Novel Still Image Coding Algorithm Based on Integer Square Threshold and Index Table

零树编码对一般小波变换有很多优点 ,但对能量集中性较差的整型小波变换 (IWT)却达不到最佳效果 针对IWT存在的这一特点 ,结合零树编码的优点给出一种新算法 第 1,根据IWT中各子带系数幅值的动态变化范围较小 ,小波图像能量集中性较DWT差这一特点 ,提出利用连续的正整数平方作为IWT小波系数量化阈值的整数平方算法 ;第 2 ,提出一种基于索引表和游程编码的小波零树编码方法 ,简化了编码与解码过程 实验表明 。

The zero tree coding algorithm has many advantages for wavelet transform, but its ability for integer wavelet transform does not equal to its ambition because the energy concentration of integer wavelet transform is not as good as wavelet transform A new algorithm is described, that improves embedded zero tree coding for integer wavelet transform from two aspects First, because every subband coefficient by integer wavelet transform has smaller dynamic change value and this kind of wavelet images have worse energy compaction than other wavelet image, a new method of selecting quantified threshold based on “integer square algorithm” is applied Second, a kind of index table based on RLE coding can help zero tree coding for better compressing images A plenty of experiments indicate that the new algorithm for IWT is more simple and efficient than the EZW algorithm and SPIHT algorithm