Image denoising with gradient projection
Yiping Chen, Xiang Li · 2011
Image denoising can be modeled as a minimization problem of L2 norm. This problem mostly is solved using L2-norm methods previously. However, L2-norm methods may smooth the results. In this paper, this problem is solved efficiently by gradient projection, in particular, minimization with L1-norm penalty as we proposed. A variable splitting technique is employed to make the L1 norm penalty function differentiable. We present a L1-norm gradient projection approach to image denoising problem where the denoising is subject to minimization with nonnegative constraints. Numerical experiments and comparisons demonstrate the effectiveness of the proposed approach.