Underwater Image Enhancement Methods Based on CNN-PDE

摘要

This paper compares Cellular Neural Network (CNN) applications of several different Partial Differential Equations (PDEs) for underwater image enhancement, including thermal diffusion equation, Laplace equation, Poisson equation, the P-M model and the TV model. The results show that the TV model can effectively preserve edge detail, and the Poisson equation is closer to the water degradation model; both of that can get a sharper image than other partial differential equations.