Modified Finite Difference and Linear Iterative PDE Methods in Digital Image Inpainting

Abstract

Image inpainting process is used to develop the damaged image or missing part of the image. This technique has more applications, such as text removal in the image, photo restoration and etc. There are different methods used in image inpainting, such as nonlinear partial differential equations, wavelet transformation, framelet transformation, etc. In this study a linear diffusion PDE method for image inpainting is considered. And to solve this linear PDE a numerical method was developed. Also, different diffusion conductivity, such as constant and nonconstant, were considered for this method. Linear diffusion PDE method was compared with existing non-linear diffusion PDE methods. For an any inpainting method, there exists an error associated with it. So, two different methods were considered to find a relationship between error and inpainting domain.