设计了一个改进算法,它仅需对小波分解矩阵进行一遍扫描即可确定层次树集合分划编码(SPIHT)算法所需的所有D(i,j)和L(i,j)重要性并完成对所有(i,j)子系数的编码,使得按照SPIHT的排序方法编码时只需查找存放D(i,j)和L(i,j)的重要性及(i,j)子系数的编码表,从而大大提高了SPIHT的编码速度。
This paper discovered an improved algorithm that requires scanning the wavelet decomposition matrix of an image only once to decide the significance of all D( i ,j) and L(i,j) and the codes of the sons of all (i,j) which were needed during the execution of SPIHT, then the coding using SPIHT' s sorting method needed only to look up the table storing these significance values and codes. Hence it could improve greatly the coding speed of SPIHT.