An Adaptive Regularization Method for Image Denoising

Abstract

This pepper proposes a new regularization method to solve the problem of low rank optimization. Approximation compared with the nuclear norm, which is a rank regularization of successive approximation. The advantage of this model is that it can directly solve the rank of the regularized problem, and it only needs to calculate a singular value decomposition to improve the efficiency. Based on Morozov discrepancy principle, this paper analyses the choosing standard of parameters and proposes an adaptive algorithm. Finally, a series of experiments were done to illustrate this algorithm. Experimental results show that the proposed algorithm is compared with other algorithms, and the algorithm not only improves the denoising effect, but also shortens the time of operation.