改进的多分辨率SPIHT算法

Improved Multi-resolution SPIHT Algorithm

由于引入了小波树,隐藏了扫描路径,因此SPIHT算法能获得较高压缩比,同时保持较好的图像解码质量。而多分辨率SPIHT算法能根据接收方的分辨率需求,使解码器根据不同信道条件选择图像还原分辨率。但是,该算法按照分辨率级成组扫描处理每级LIP,LIS,LSP表,更新下一级表时,会造成重复比较和冗余编码,既浪费执行时间,又增加了计算复杂度。该文改进了该算法,简化了原有算法流程,减少了编码冗余。理论分析和实验表明,在保持较高PSNR的同时,该算法明显提高了编码速度。

Since the introduction of wavelet tree and certain pass, SPIHT can achieve high compression ratio as well as high image quality. Later multi-resolution SPIHT is proposed to enable the decoder to select certain resolution as channel condition permits. However, the original algorithm produces repeated comparison and redundant coding in resolution-based sorting, wasting time and increasing complexity when sorting LIP, LIS LSP, tables while updating them at next level. In the paper, a reduced algorithm is introduced to decrease comparison time so as to promote compression rate with high PSNR. Theoretical and simulation analysis indicates that the proposed method can promote the encode speed and the PSNR can be guaranteed.