基于ARM9的整数小波阈值压缩算法及其EVC实现

Integer wavelet threshold compression algorithm based on ARM9 and its EVC realization

采用传统的一代小波变换进行数据压缩,存在实时性差,内存需求量大等问题。本文从小波滤波器的多相位矩阵出发,利用Euclidean分解方法,实现了小波变换由传统卷积模式到提升算法的转变,并进行了整数提升,克服了一代小波的不足。在此基础上提出了一种改进的整数小波阈值压缩算法,在ARM9上利用EVC予以实现。实验表明该算法在均方误差为0.3490%时,压缩比为11.875%,计算速度快,适合应用在实时性要求较强的场合。

When using traditional first generational wavelet transform for data compression, the algorithm had bad real-time character and great memory demand. Using polyphase matrix of wavelet filter and method of Euclidean analysis,the transformation of wavelet from tradi- tional convolution model to lifting scheme was realized, then integer lifting was performed in this paper, which overcame the limitation caused by the first generation wavelet transform. Based on the above, an improved integer wavelet threshold compression algorithm was proposed and realized on ARM9 using EVC. The experimental results show that when mean square error is 0.3490%, the data compression ratio of our proposed algorithm can be 11.875% and the speed is fast, so it is suitable to the eases where the real-time data eompression is required.