摘要
作为JPEG2000的重要组成部分,整数小波变换(IntegerWaveletTransform)和嵌入式零树编码(EmbeddedZerotreeWaveletCoding)有许多优越性,但是图象经整数小波变换(IWT)后各子带系数幅值的动态变化较小,因此其能量集中性较第一代小波变换差很多,不利于零树编码(EZW)。该文从零树编码的量化阈值入手,将传统零树编码中“2的整数次幂”量化阈值改为从1开始的整数平方量化阈值,通过缩短了各量化阈值间的距离,减少了重要系数集中出现在较低量化阈值的机会,增加了编码过程中零树的数量,从而充分利用了整数小波变换能量集中性较低的特点。实验证明,该算法通过整数平方量化阈值有效地将整数小波变换同零树编码结合起来,提高了编码效率。
Integer wavelet transform(IWT )and embedded zerotree wavelet(EZW)coding are the important components of JPEG2000because of their many advantages.But every subband coefficients of the wavelet image by IWT have smaller dynamic change value,so this kind of wavelets have worse energy compaction than the first generation wavelet and do not take full advantage of EZW coding.In this paper,a new algorithm is introduced with quantization threshold of EZW.It makes use of'Integer Quantization Square Threshold From1'instead of'2of integer power'.It shortens the dis-tances of every quantization threshold,decreases the probability that the important coefficients concentrate on the lower quantization threshold area,improves the quantity of zerotree and takes full advantage of the worse energy compaction of IWT.Experiments prove that this algorithm makes full use of EZW and IWT by'Integer Quantization Square Threshold'and improves the efficiency of coding.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第1期35-38,共4页
Computer Engineering and Applications
基金
国家自然科学基金"城市交通流诱导系统理论模型和方法的研究"(编号:59638220)的子课题"交通流中的高速图象压缩传输"的资助
