文章导读
总览 评价 曹坤 , 徐文波 * , 郭莉 ( 北京邮电大学信息与通信工程学院,北京 100876; ) 摘要: 重建算法是压缩感知理论研究中的一个重要部分。本文基于光滑的L0范数和子空间追踪法提出一种新的压缩感知重建算法---SL0SP(Subspace Pursuit base on Smoot
曹坤, 徐文波*, 郭莉
(
北京邮电大学信息与通信工程学院,北京 100876; )
摘要:
重建算法是压缩感知理论研究中的一个重要部分。本文基于光滑的L0范数和子空间追踪法提出一种新的压缩感知重建算法---SL0SP(Subspace Pursuit base on Smoothed L0)。利用光滑函数逼近L0范数,得到一种新的最优化问题,并采用梯度法进行初步求解,在此基础上利用子空间追踪法得到最后的重建信号。实验结果表明,在相同的测试条件下,SL0SP算法能够得到比SP(Subspace Pursuit)算法更好的重建效果。
关键词:
压缩感知;稀疏重建;光滑L0范数;子空间追踪法
CAO Kun, XU Wenbo*, GUO Li
(
School of Information and Communication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876; )
Abstract:
Approach for recovery algorithm is an important part of compressive sensing (CS) theroy. In this paper, we propose a new reconstruction algorithm based on smoothed L0 norm and subspace pursuit (SP), termed as the SL0SP algorithm. We employ a suitable continuous function to approximate the discontinuous function L0 norm, yielding a new optimization problem, and get an initial solution by steepest descent method, then utilize SP method to reconstruct the signal. Experimental resuts show that the proposed SL0SP algorithm is superior to existing SP method in terms of reconstruction quality.
Tag:
点此返回栏目查看更多>>>参考论文