摘要
针对压缩感知理论中迭代硬阈值IHT重构算法要求给定信号稀疏度的缺点,提出了一种变步长稀疏自适应迭代硬阈值VSSSAIHT算法。该算法在信号的稀疏度未知的情况下,通过相邻迭代残差的差值大小来选择合适的步长,以扩大重构信号的支撑集,不断逼近原始信号的稀疏度,逐步迭代恢复信号。仿真结果表明,与迭代硬阈值算法相比,VSSSAIHT算法改善了图像重构的质量,减少了算法运行的时间。
Aiming at the shortcomings that iterative hard thresholding（IHT）reconstruction algorithm of compressed sensing theory requires the sparsity of original signal is known, a variable step size sparsi- ty adaptive iterative hard thresholding（VSSSAIHT）algorithm was proposed. When the sparsity of origi- nal signal is unknown, the proposed algorithm according to the differences between adjacent residuals chooses appropriate step size to increase the number of support set of the reconstructed signal, approxi- mate the sparsity of original signal gradually and restore signals by gradual iterations. Simulation results show that the VSSSAIHT algorithm, compared with the IHT algorithm, improves the quality of the re- constructed image, and reduces running time.
出处
《计算机工程与科学》
CSCD
北大核心
2013年第8期120-124,共5页
Computer Engineering & Science
基金
国家自然科学基金资助项目(60872064)
关键词
压缩感知
迭代硬阈值
图像重构
变步长
稀疏自适应
compressed sensing
iterative hard thresholding
image reconstruction
variable step size
sparsity adaptive
