ADMM算法在3D彩色图像恢复中的应用

Application of ADMM Algorithm in 3D Color Image Recovery

本文讨论稀疏信号的恢复和分离问题。利用加权l1范数进行稀疏诱导,提出了加权l1范数极小化约束模型,利用ADMM算法进行求解。在适当地假设下证明了算法的收敛性。对带有盐噪声和椒盐噪声的3D彩色图像进行了数值实验,并与JP算法及YALL1算法进行了数值比对。实验结果的峰值信噪比PSNR和相对误差RelErr表明,我们的算法具有较好的恢复效果。

This paper discusses the recovery and demixing problem of sparse signals. We propose a weighted l1 norm minimization model. Then the ADMM algorithm is applied to this model. We also prove the convergence property of our algorithm under mild conditions. At last, we do two numerical experiments for 3D color image, in which salt noise and salt-and-pepper noise are chosen. In addition, we compare our algorithm with two other algorithms: the JP and the YALL1. Numerical results of PSNR and RelErr show that our algorithm gives relatively better behavior.