基于块稀疏度与自适应迭代的压缩感知方法

A Compressed Sensing Method Based on Block Sparsity and Adaptive Iteration

针对分块压缩感知算法在平滑块效应时损失了大量的细节纹理信息,从而影响图像的重构效果问题,提出了一种基于块稀疏信号的压缩感知重构算法。该算法先采用块稀疏度估计对信号的稀疏性做初步估计,通过对块稀疏度进行估算初始化阶段长,运用块矩阵与残差信号最匹配原则来选取支撑块,再运用自适应迭代计算实现对块稀疏信号的重构,较好地解决了浪费存储资源和计算量大的问题。实验结果表明,相比常用压缩感知方法,所提算法能明显减少运算时间,且能有效提高图像重构效果。

For the problem that block compressed sensing algorithm loses a lot of detail texture information when smoothing block effect,which affects the image reconstruction effect,a compressed sensing reconstruction algorithm based on block sparse signal is proposed.Firstly,the sparsity of the signal is estimated by block sparsity estimation method,and then the step size is initialized.The support block is selected by using the principle of the best match between block matrix and residual signal.Then,the adaptive iteration calculation is performed to reconstruct the sparse signal,which solves the problem of wasting storage resources and large amount of computation.The experimental results show that the proposed algorithm can significantly reduce the computational time and effectively improve the image reconstruction effect compared with the commonly used compressed sensing methods.