首页 | 本学科首页   官方微博 | 高级检索  
     

基于FFT的MP信号稀疏分解算法的改进
引用本文:邵君,尹忠科,王建英. 基于FFT的MP信号稀疏分解算法的改进[J]. 西南交通大学学报, 2006, 41(4): 466-470
作者姓名:邵君  尹忠科  王建英
作者单位:西南交通大学信息科学与技术学院,四川,成都,610031
基金项目:国家留学基金资助项目(21851039);教育部留学回国人员科研启动基金资助项目(教外司[2004]517号);四川省应用基础研究项目(03JY029-048-2;04JY029-059-2)
摘    要:针对基于FFT的MP信号稀疏分解算法中存在的计算量过大的问题,提出了改进算法.改进算法充分利用了当FFT算法的变换长度是2的整数次幂时运算速度最快的性质,用基2 FFF实现信号稀疏分解中的相关运算.理论分析显示,当数字信号长度为1 024采样点时,用FFT算法计算互相关的速度为直接计算的10.6倍.仿真实验结果表明,改进算法的计算速度为直接计算的8.05倍,为原基于FFT的MP算法的3.64倍.

关 键 词:信号处理 稀疏表示 稀疏分解 MP算法 快速傅立叶变换 改进
文章编号:0258-2724(2006)04-0466-05
收稿时间:2005-04-11
修稿时间:2005-04-11

Improved FFT-Based MP Algorithm for Signal Sparse Decomposition
SHAO Jun,YIN Zhongke,WANG Jianying. Improved FFT-Based MP Algorithm for Signal Sparse Decomposition[J]. Journal of Southwest Jiaotong University, 2006, 41(4): 466-470
Authors:SHAO Jun  YIN Zhongke  WANG Jianying
Affiliation:School of Information Science and Tech. , Southwest Jiaotong University, Chengdu 610031, China
Abstract:An improved FFT(fast Fourier transformation) based MP(matching pursuit) algorithm was proposed to reduce the calculation load in signal sparse decomposition.In the algorithm,a radix-2 FFT is performed to calculate the crosscorrelation, because the calculation speed is the fastest when the length of the transform is of integral power of 2.Theoretical analysis shows that,for a digital signal with a length of 1 024,the speed of calculation of crosscorrelation using radix-2 FFT is 10.6 times as fast as that using direct calculation.Simulation results show that the calculation speed of the improved FFT based MP algorithm is 8.05 times as fast as that of direct calculation,and 3.64 times as fast as that of the FFT-based MP algorithm.
Keywords:signal processing   sparse representation   sparse decomposition   MP algorithm   FFT   improvement
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号