Novel explanation, modeling and realization of Lattice Boltzmann methods for image processing

Abstract

Image processing is a cost-effective technology in various applications. Partial differential equation (PDE) methods are popular when realizing image processing. However, computational speed of existing PDE methods cannot meet requirements in practice. To solve this problem, scholars proposed a novel method: Lattice Boltzmann (LB) model. Although LB model has already been applied for image denoising, inpainting and segmentation, its explanation is not systematically concluded and a general LB model for image processing is missing, which resulted in previous investigations difficult to be scaled up. The purpose of this paper is to explore the explanation of LB model for image processing, and propose a general LB mathematical model. To test the feasibility of the proposed LB model, we did several comparison experiments. The comparison results showed that the proposed LB model augmented CPU calculating speed and kept good image processing effect.