摘要
在对M a llat算法结构进行改进的同时,将长序列快速卷积算法中的重叠保留法引入M a llat算法中,提出了一种适合长序列小波变换的快速算法,给出了数学推导过程和具体实现步骤。该方法大大降低了小波变换的计算量,且并行性很好。仿真实验结果验证了算法的正确性,且运算速度较直接线性卷积实现方法有很大提高。
Improved algorithm structures based on the Mallat algorithm are designed, and the overlap-saving method for the computation of the long sequence fast convolution is applied in the Mallat algorithm. An improved fast wavelet transform algorithm suitable for the long sequence is proposed and its theoretical derivation and detailed implementation steps are given. The method reduces the computation amount of the wavelet transform and has a good parallel property. Simulation results prove that the proposed algorithm is accurate. And the computation efficiency is improved compared with the traditional linear convolution.
出处
《数据采集与处理》
CSCD
北大核心
2006年第1期86-89,共4页
Journal of Data Acquisition and Processing
基金
国防预研基金资助项目