项目名称: 变分法和偏微分方程理论在图像重建中的应用
项目编号: No.11201341
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 数理科学和化学
项目作者: 王伟
作者单位: 同济大学
项目金额: 20万元
中文摘要: 近年来, 变分法和偏微分方程在图像处理中的应用越来越广泛, 得到了许多令人鼓舞的结果。 本项目主要应用变分法和偏微分方程理论处理各种图像重建问题。 主要研究内容包括: 图像去噪、 图像去模糊、 图像色彩恢复、 高分辨率图像及视频重建, 三维全景图像生成及编辑。 主要研究思路是在传统图像处理的框架下引进变分法及偏微分方程理论, 提出一些新的变分模型, 并结合如算子分裂方法、 辅助变量法、 ADMM算法、 Bregman迭代法等高效算法求解极小解。 理论研究包括: 能量泛函极小解的存在唯一性, 相应欧拉-拉格朗日方程及热流方程解的存在唯一性、 稳定性、 正则性, 数值算法的收敛性等。 本项目的研究可以从模糊的、 分辨率低的图像或者视频得到清晰的、 高分辨率的图像或者视频, 对于图像处理的发展具有重要意义。
中文关键词: 变分法;图像重建;算法;迭代;
英文摘要: The applications of variational methods and partial differential equations have been proved to be very effective in image processing. Many results have been shown to be very encouraging in this area. In this project, we focus on dealing with various image reconstruction problems by using variational methods and partial differential equations. The main content includes: image denoising, image deblurring, color image restoration, high resolution image and video reconstruction, 3D image stitching and editing. Specifically, we will propose some new variational models based on different purpose by introducing some theory of PDE into image processing. In order to solve the minimization problems, we will apply some efficient methods such as operator splitting method, auxiliary variable method, ADMM method and Bregman iteration. The theory research includes: existence and uniqueness of the minimization problems; existence, uniqueness, stability, and regularity of the Euler-Lagrange equations or the corresponding Heat equations. This project is very important for the development of image processing because we can derive clear and high resolution images or video from blurry and low resolution images or video.
英文关键词: Variational approach;Image reconstruction;Algorithm;Iteration;