Poisson Noise Removal Scheme Based on Fourth-Order PDE by Alternating Minimization Algorithm

Abstract

To overcome the staircasing effects introduced by the TV regularization in image restoration, this paper investigates a fourth-order partial differential equation (PDE) filter for removing Poisson noise. In consideration of the slow convergence property of the classical gradient descent method, we adopt the alternating minimization algorithm for realizing this scheme. Compared with the corresponding total variation based one, numerical simulations distinctly indicate the superiority of our proposed strategy in handling smooth regions of Poissonian images and improving the computational speed.