Abstract: The Sparsity Adaptive Matching Pursuit（SAMP） algorithm has a large range of application in compressive sensing, but it runs slowly and the performance of recovery is not good. Compared with SAMP, a novel adaptive subspace pursuit algorithm is presented, which uses the idea of stage, evaluates the sparsity of the original signal step by step, and then with the information of sparsity, recovers the original signal using the subspace pursuit algorithm. The experiments demonstrate that the new algorithm not only improves the performance of the recovery, and saves the operating time compared with SAMP, but also solves the problem of unknown sparsity K in SP.

Key words: compressive sensing, signal recovery, adaptive, subspace pursuit

摘要： 针对稀疏自适应匹配追踪（SAMP）算法中存在的运行速度慢、重建效果欠佳的问题，提出了一种新的自适应的子空间追踪算法（MASP）。采用SAMP算法中分段的思想，先对半减小预估稀疏度，再逐一增加，得到真实稀疏度后，再利用子空间追踪算法对原始信号进行重构。实验表明，相比于SAMP算法，该算法在相同观测数量的情况下，具有较快的运行时间和较好的重建效果，其中，在重构信噪比方面平均提高8.2%。

关键词: 压缩感知, 信号重构, 自适应, 子空间追踪