基于流体力学的图像修复.doc

安徽大学硕士毕业论文 基于流体力学的图像修复 何信息，所以也不能满足“连接性准则”，因此我们对该模型进行了改进，在修 复方程中添加了曲率驱动的扩散项，它使改进后的修复方法能够满足“连接性 准则”，最后通过模拟实验证明该改进方法的有效性。 关键词：图像修复、偏微分方程、流体力学、Navier-Stokes、TV-Stokes II Abstract Abstract Image inpainting is the process of filling in missing data in a designated region of a still or video image. The goal is to produce a revised image in which the inpainted region is seamlessly merged into the image in a way that is not detectable by a typical viewer. Image inpainting is not only an important part of image restoration research, but also a hotspot in the field of Image Processing and Computer Vision research. The paper is to study the digital image inpainting, that is, we have to use the computer to inpaint the damaged image. Image can be simply divided into the structure image and the texture image two categories, so image inpainting can be accordingly divided into the structure image inpainting and the texture image inpainting two categories. The basic method of texture image inpainting is the texture synthesis, which is a specialized research. The structure image mainly uses PDE-based image inpainting models. PDE-based image inpainting models main ideal is to establish set of partial differential equations according to energy minimization mode1s, and consider it as a boundary problem of partial differential equations, and then solve it by time match method. Because it can accurately build model for images, so it has solved well many complex problems of image inpainting. In the paper, we mainly study several types of PDE-based image inpainting models: Total Variation (TV) inpainting model, Curvature Driven Diffusion (CDD) model，which improves on TV model. Then, we discuss in detail about two new PDE-based image inpainting models, that is, the image inpainting model using a Navier-Stokes equation and the image inpainting model using a TV-Stokes equation. They compare the image intensity to a fluid. The image inpainting model using a Navier-Stokes equation imitates the Na