Abstract: Aiming at the problems of worse reconstructed image quality and larger time complexity of image Compressed Sensing(CS) reconstruction algorithms, this paper presents a Double Shrinkage Fast Iterative Algorithm(DSFIA). This algorithm, which brings in threshold shrinkage and regularization parameter shrinkage, restores image signals by gradual iteration in order to speed up the convergence rate and improve the quality of restored images. Simulation results show that, compared with traditional threshold iterative algorithms, it makes Peak Signal to Noise Ratio(PSNR) of the reconstructed image be higher, and reduces running time when the sampling rate is low.

Key words: Compressed Sensing(CS), image reconstruction, threshold shrinkag, Fast Iterative Shrinkage-thresholding Algorithm(FISTA), Double Shrinkage Fast Iterative Algorithm(DSFIA), regularization parameter shrinkage

摘要： 针对传统图像压缩感知重构算法重构质量差及时间复杂度大的问题，提出一种双收缩快速迭代算法。通过引入阈值和正则化参数的双收缩，逐步迭代恢复图像信号，以加快收敛速度，改善重构质量。仿真结果表明，与传统阈值迭代算法相比，该算法重构图像的峰值信噪比较高，在低采样率下运行时间较少。

关键词: 压缩感知, 图像重构, 阈值收缩, 快速迭代收缩阈值算法, 双收缩快速迭代算法, 正则化参数收缩

CLC Number: 

• TN911.73