摘要
压缩感知理论,从信号的自身特性出发,通过变换作用域和线性投影实现对信号的采样和压缩。测量矩阵是该理论中获得的最优测量,实现精确重构的关键。本文在介绍常用测量矩阵的基础上,重点研究了结构化测量矩阵。鉴于测量矩阵设计的最重要的原则是降低矩阵元素间的相干性,借鉴循环矩阵和广义轮换矩阵的优点,提出了采用均匀随机数对结构化测量矩阵进行随机循环的构造方法。仿真实验表明新矩阵在信号重建上具有更好的性能。
Compressed sensing, depending on the fact that signals are sparsely or compressive in some fixed basis, samples and compresses the signal by linear projection. Measurement matrix plays the positive role in the theory. The common measurement matrix is researched, especially the structured matrix. As minimizing the coherence is an important way to design measurement ma- trix, the advantages of circulant matrix and generalized rotation matrix are learned and a new method to design a measurement matrix which based on random circulant is proposed. The simulation results suggest the new matrix has better performance in signal' s reconstruction.
出处
《电视技术》
北大核心
2015年第19期6-9,共4页
Video Engineering
关键词
压缩感知
测量矩阵
托普利兹矩阵
随机循环
compressed sensing
measurement matrix
Toeplitz matrix
random circulant