基于提升小波技术的阈值量化图像压缩方案

Image Compression Quantification Scheme Based on Lifting Wavelet Technology

结合多级树集合分裂算法(SPIHT)能产生完全的嵌入式码流的特性,设计了一种新的阈值量化方法的彩色图像压缩算法。通过建立彩色空间的可逆分量变换,减小三个彩色分量之间的相关性,并对变换后的三分量进行提升的整数小波变换。将定义的自然数的平方作为门限值,完成对三分量小波系数矩阵的量化和联合编码。实验结果表明,本方案与传统量化方案下的SPIHT算法相比,大多数重要系数的编码长度至少减少1个比特数,当比特率非常低时,压缩效果更明显。

Combining with the characteristic of set partitioning in hierarchical trees （SPIHT）, which could produce completely embedded code stream, a novel quantification method for color image compression was designed. By establishing the reversible component transformation （RCT） in color space, the three-component was transformed with less relevance, and then after the lifting integer wavelet transformation, the processed components were still integers. Based on the theory of SPIHT, the proposed method set the square of defined integers as thresholds, completed the quantization and united coding of the three-component wavelet coefficients in the same time. The experiment results show that most important wavelet coefficients need less coding bits, compared with the traditional quantification, each one at least has shorten one bit. In the case of lower transmission rate, the compression effect shows better.