Computer Engineering and Applications ›› 2016, Vol. 52 ›› Issue (5): 79-83.

Previous Articles     Next Articles

Orthogonal Sem-Iterative Hard Threshold (OSIHT) reconstruction for compressed sensing

YANG Aiping, LIU Huaping, HE Yuqing, LI Gai   

  1. School of Electronic Information Engineering, Tianjin University, Tianjin 300072, China
  • Online:2016-03-01 Published:2016-03-17

正交半迭代硬阈值(OSIHT)压缩感知重构算法

杨爱萍,刘华平,何宇清,栗  改   

  1. 天津大学 电子信息工程学院,天津 300072

Abstract: Signal reconstruction plays an important role in the compressed sensing process. Iterative Hard Threshold (IHT) algorithm has found a wide application in signal reconstruction due to its good performance, but it also shows very low convergence rate. The Semi-Iterative Hard Thresholding (SIHT) algorithm recently proposed can achieve fast convergence, but it is sensitive to the scaling of the measurement matrix and has a strong dependence on measurement matrix, which greatly limits its application. Inspired by the idea that OMP improves MP algorithm, and in order to improve the SIHT algorithm, the Orthogonal Semi-Iterative Hard Threshold (OSIHT) algorithm is put forward. The proposed algorithm not only removes the restriction on the orthogonal measurement matrix, but significantly improves the quality of reconstructed image, while greatly shortening the operation time.

Key words: compressed sensing, reconstruction algorithms, iterative threshold, semi-iterative method

摘要: 信号重构是压缩感知过程中的重要环节,迭代硬阈值(IHT)算法因具有较好的重构性能被广泛应用,但其收敛速度比较慢。近期提出的半迭代硬阈值算法(SIHT)虽然可实现快速收敛,但对测量矩阵的尺度缩放非常敏感,依赖性强,大大限制了其应用范围。受OMP对MP算法改进启发,对SIHT算法进行改进,提出了正交半迭代硬阈值(OSIHT)重构算法。该算法不仅取消了对测量矩阵的依赖性,还有效改善了图像重构质量,减少运行时间。

关键词: 压缩感知, 重构算法, 迭代阈值, 半迭代法