论文部分内容阅读
从条件概率的角度分析了压缩感知信号重构,认为重构信号是在观测信号发生的条件下所发生的后验信号.基于信号的条件概率分析,提出一种简单而有效的算法优化方法,此方法可以提高贪婪追踪系列算法选择原子的准确性,在某种程度上可以削减感知矩阵自身相关度的影响.此外,本文概述了系列贪婪追踪算法,并应用本文提出的方法来改进它们.模拟实验表明,本文提出的方法可以改进大多数贪婪追踪算法,使算法在速度和精度上都有所提高.
Compressed sensing signal reconstruction is analyzed from the view of conditional probability, and it is considered that the reconstructed signal is a posterior signal that occurs under the condition of observed signal.A simple and effective algorithm optimization method is proposed based on conditional probability analysis of signal, This method can improve the accuracy of selecting atom for the algorithm of greedy tracking, and to a certain extent, reduce the influence of the correlation matrix of the sensing matrix.In addition, this paper outlines a series of greedy tracking algorithms, and applies the method proposed in this paper to improve them. Experiments show that the proposed method can improve most of the greedy tracking algorithm, the speed and accuracy of the algorithm are improved.