A novel adaptive image zooming method based on nonlocal Cahn–Hilliard equation

Abstract

Recently, nonlocal operator has become a very hot topic in image processing. For nonlocal partial differential equations (PDEs) methods, almost existed models are based on second order. In this paper, a novel image zooming method based on nonlocal fourth-order PDEs is proposed. The proposed method comprehensively utilizes the characteristic of the higher order PDEs-Cahn–Hilliard equation and the nonlocal operator. Higher order PDEs can well alleviate blocking artifacts and decrease the influence of the noise, and the nonlocal operator has a good performance on texture preserving and can reduce Gaussian noise. In the proposed model, a control parameter is introduced, which can control the speed of the model diffusion. Experimental results show that the proposed method has a good ability for de-noising and texture preserving.