基于块稀疏度估计的压缩感知自适应重构算法

摘要

The block-sparse signal is a kind of sparse signal with special structure.In the case of the block sparsity is unknown,this paper proposed an adaptive reconstruction algorithm based on estimating of block sparsity in compressed sensing.Firstly,the algorithm obtained an estimated value of support index set by preliminary estimates calculation of the block sparsity,and then it initialized the residual with the estimated value.Next,it obtained a support set which calculated by the correlation matching operations and determined by the regularization filter.Finally,it acquired the correct support set by iteration process.Simulation results show that the proposed algorithm is with better recovery probability and shorter average running time compared with the existing adaptive reconstruction algorithm for block-sparse signal.%块稀疏信号是一类具有特殊结构的稀疏信号.针对块稀疏信号块稀疏度未知的情况,提出了一种基于块稀疏度估计的自适应重构算法,并将其应用于压缩感知.首先对信号的块稀疏度进行初步估计,计算得到一个支撑块索引集合的估计值,利用得到的估计值对残差进行初始化;然后对测量矩阵的子块和当前残差进行相关性匹配操作,以选取信号的支撑块集合,依据正则化原则再次对由相关性匹配操作得到的信号支撑块集合进行筛选;最后通过迭代过程获得信号最终的支撑块集合.仿真实验结果表明,提出的算法与现有的块稀疏信号自适应重构算法相比,具有较好的重构成功概率且算法的平均运行时间更短.