摘要：

压缩感知重构算法在实际应用中需要预知信号稀疏度，而信号的稀疏度通常是未知的。为此，改进压缩采样匹配追踪(CoSaMP)算法的自适应性，提出一种稀疏度自适应贪婪算法。对信号稀疏度进行初始估计，结合SAMP算法思想，以残差值比对为终止条件，在CoSaMP算法框架下进行稀疏度逐步增大的递归运算，实现精确重构。仿真实验结果证明，该算法重构精度高、抗噪能力强，同时具备稀疏度自适应的特点。

关键词: 压缩感知, 贪婪重构算法, 匹配追踪, 递归, 稀疏估计

Abstract:

Compressive sensing reconstruction algorithm needs to predict sparse degree of the signal, however the sparse degree is always impossible to obtain in the practical application. In order to improve the sparsity adaptive problem of the CoSaMP algorithm, a new algorithm is proposed for the signal reconstruction. It initially estimates the sparse degree, combined with the idea of SAMP algorithm, the sparse degree gradually increases under the framework of CoSaMP algorithm. With the comparison of residual value for termination condition, it can exactly recover original signal after few recursive operations. Results show that the new algorithm can exactly recover original signal at the noisy condition and owns the characteristics of sparsity adaption.

Key words: compressive sensing, greedy reconstruction algorithm, matching pursuit, recursion, sparse estimation

中图分类号: 

• TN919