关键词: 图像去噪, 非线性总变差模型, 加性算子分裂

Abstract: An addtive operator splitting (AOS) method is employed in the nonlinear total variation model (ROF model) for image denoising. The numerical discretization scheme is semiimplicit and unconditional stable. It overcomes the time step size restriction for explicit scheme. The method can easily be parallelized and one can gain an accelerated convergence. The experiments show that it is extremely promising compared to the explicit scheme.

Key words: image denoising, nonlinear total variation model, additive operator splitting