北京航空航天大学学报 ›› 2017, Vol. 43 ›› Issue (10): 2109-2117.doi: 10.13700/j.bh.1001-5965.2016.0830

• 论文 • 上一篇    下一篇

基于CS的正则化稀疏度变步长自适应匹配追踪算法

刘浩强1, 赵洪博2, 冯文全2   

  1. 1. 北京航空航天大学 大型飞机高级人才培训班, 北京 100083;
    2. 北京航空航天大学 电子信息工程学院, 北京 100083
  • 收稿日期:2016-10-27 修回日期:2016-12-23 出版日期:2017-10-20 发布日期:2017-02-21
  • 通讯作者: 赵洪博 E-mail:bhzhb@126.com
  • 作者简介:刘浩强,男,硕士研究生。主要研究方向:信号处理、压缩感知、室内定位和信息融合技术;赵洪博,男,博士,讲师,硕士生导师。主要研究方向:卫星导航、飞行器通信与测控等相关理论和关键技术;冯文全,男,博士,教授,博士生导师。主要研究方向:卫星通信与测控、卫星综合测试与仿真、卫星导航。
  • 基金资助:
    国家自然科学基金(91438116);中国航天科技创新基金(2016-1-107)

Regularized sparsity variable step-size adaptive matching pursuit algorithm for compressed sensing

LIU Haoqiang1, ZHAO Hongbo2, FENG Wenquan2   

  1. 1. Large Aircraft Advanced Training Class, Beijing University of Aeronautics and Astronautics, Beijing 100083, China;
    2. School of Electronic and Information Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
  • Received:2016-10-27 Revised:2016-12-23 Online:2017-10-20 Published:2017-02-21

摘要: 压缩感知(CS)能够突破Nyquist采样定理的瓶颈,使得高分辨率信号采集成为可能。重构算法是压缩感知中最为关键的部分,迭代贪婪算法是其中比较重要的研究方向。对压缩感知理论进行了详细分析,并在现有重构算法的基础上提出了一种新的迭代贪婪算法——正则化稀疏度变步长自适应匹配追踪(RSVssAMP)算法,可在信号稀疏度未知的情况下,结合正则化和步长自适应变化思想,快速精确地进行重构。相比于传统迭代贪婪算法,本文算法不依赖于信号稀疏度,并且应用正则化以确保选取支撑集的正确性。此外,应用自适应变化步长代替固定步长,能够提高重构速率,而且达到更高的精度。为了验证本文算法的正确性,选取高斯稀疏信号和离散稀疏信号分别进行仿真,并与现有算法进行比较。仿真结果表明,本文算法相比于现有算法可以实现更加精确快速的重构。

关键词: 压缩感知(CS), 自适应, 正则化, 变步长, 匹配追踪

Abstract: Compressed sensing (CS), which could break through the bottleneck of the Nyquist sampling theorem, makes the high resolution signal acquisition possible. Reconstruction algorithm is the key part of compressed sensing, and the iterative greedy algorithm is one of highly significant research directions. A novel iterative greedy algorithm for compressed sensing, named regularized sparsity variable step-size adaptive matching pursuit (RSVssAMP) algorithm, was proposed in this paper. The regularized idea and the variable step-size adaptive idea were utilized in the new algorithm to achieve a quick and accurate reconstruction under the condition that the sparsity of a signal was unknown. Compared with traditional greedy algorithms, RSVssAMP could reconstruct the signal without prior information of the sparsity, and it could accelerate the reconstruction speed obviously and achieve better performance by acquiring a better candidate set. The Gaussian sparse signal and discrete sparse signal were taken as trial signals, and the comparisons of reconstruction probability and time were demonstrated in this paper. The simulation results indicate that the proposed algorithm could achieve a higher reconstruction precision and take shorter time when compared with the existing greedy algorithms.

Key words: compressed sensing (CS), adaptive, regularized, variable step-size, matching pursuit

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发