改进的压缩感知重构算法研究

Research on Improved Compressive Sensing Reconstruction Algorithm

为了对未知稀疏度信号、特殊信号、含噪声信号进行准确重构,提出一种改进的压缩感知重构算法——预测正交匹配追踪算法。提出的算法通过所选支撑集内原子总数、信号间能量差以及残差共同预测并选择所需原子。预测正交匹配追踪算法能够在稀疏度未知的情况下自适应地对块稀疏信号、噪声信号及图片信号进行准确重构。实验结果表明,在相同条件下,改进后的算法提高了重构质量,减少运行时间。

In order to accurately reconstruct the unknown sparse signal,the special signal and the noise signal,an improved compressive sensing reconstruction algorithm—a forecast orthogonal matching pursuit algorithm proposed.The proposed algorithm predicts and selects the desired atoms by the total number of atoms in the selected support set,the energy difference between the signals and the residuals information.The forecast orthogonal matching pursuit algorithm can adaptively reconstruct the block sparse signals,the noise signals and the picture signals in the case of unknown sparse degree.The experimental results show that the improved algorithm improves the quality of reconstruction and reduces the running time under the same conditions.