基于CS的正则化稀疏度变步长自适应匹配追踪算法

发布时间：2018-07-02 19:19

本文选题：压缩感知(CS) + 自适应　；参考：《北京航空航天大学学报》2017年10期

【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(RSVss AMP)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。
[Abstract]:Compression sensing (CS) can break through the bottleneck of Nyquist sampling theorem and make high resolution signal acquisition possible. The reconstruction algorithm is the most important part of compression perception, and iterative greedy algorithm is one of the most important research directions. The theory of compression perception is analyzed in detail, and a new iterative greedy algorithm, regularized sparse variable step size adaptive matching tracing (RSVss AMP) algorithm, is proposed on the basis of existing reconstruction algorithms, which can be used in the case of unknown signal sparsity. Combined with the idea of regularization and adaptive change of step size, the reconstruction is carried out quickly and accurately. Compared with the traditional iterative greedy algorithm, the algorithm does not depend on the signal sparsity and applies regularization to ensure the correctness of the selection support set. In addition, using adaptive variable step instead of fixed step can improve the reconstruction rate and achieve higher accuracy. In order to verify the correctness of this algorithm, Gao Si sparse signal and discrete sparse signal are selected for simulation, and compared with the existing algorithms. Simulation results show that the proposed algorithm is more accurate and faster than the existing algorithms.
【作者单位】：北京航空航天大学;北京航空航天大学电子信息工程学院;
【基金】：国家自然科学基金(91438116) 中国航天科技创新基金(2016-1-107)~~
【分类号】：TN911.7

【相似文献】