研究非负约束全变分图像去模糊问题,提出了一个基于增广拉格朗日方法的积极集方法,并证明了该方法在有限步内可求解,进一步推出该方法等价于解非光滑方程组的半光滑牛顿法.
In this paper we investigate the total variation image deblurring problem with the non-negativity constraint. An active set strategy based on the augmented Lagrangian method is presented and its convergence in finitely many steps is proved. Furthermore, we deduce that this active set strategy can be regarded as a semismooth Newton method for solving a system of nonsmooth equations.