项目名称: 基于微分方程模型的介质成像和图像处理的数值方法
项目编号: No.91330109
项目类型: 重大研究计划
立项/批准年度: 2014
项目学科: 数理科学和化学
项目作者: 刘继军
作者单位: 东南大学
项目金额: 70万元
中文摘要: 介质成像和图像处理的有效数值方法是现代计算数学的一个重要研究领域,数学上它可以归结为数学物理反问题和不适定问题的研究。本项目以课题组已有的反问题和图像处理的研究工作为基础,研究下面三个方面的内容:以声波和电磁波的散射为基础的散射体成像的数学模型和数值实现;以散度型椭圆方程为模型的生物介质成像的数值方法;以分数阶偏微分方程为模型的图像处理的数值方法。这三方面内容在微分方程不适定问题求解的框架下彼此关联,核心的问题都是在数据不足和噪音数据的情形下建立成像的正则化方法模型和高效数值方法。项目的研究将为介质的无损检测和目标探测、生物图像处理、实时三维人脸表情重建等图像处理领域建立合适的数学模型并提供坚实的数学分析基础,进而提供有效的数值实现方案和具体的软件实现,并将算法应用于具体的工程成像领域。依托此项目,培养一支高水平的位于国际前沿的可计算建模和科学计算的学术团队。
中文关键词: 介质成像;数学建模;不适定问题;正则化方法;数值解
英文摘要: The efficient schemes for media imaging and image processing are important research areas in modern scientific computations. From the mathematical points of view, these problems belong to the categories of inverse and ill-posed problems for mathematical physics. Based on our previous researches on inverse problems and imaging process, this project covers three research areas arising in media imaging and image processing: reconstructions for complex media based on the wave scattering models; biomedical imaging based on the elliptic equation in divergence form; and numerical schemes of image process from the PDE model of fractional order. All these problems are related each other in the framework of inverse problems for PDEs, with the key problems that the regularizing image models with efficient implementation schemes are required in the cases of insufficient and noisy input data. This project will provide appropriate mathematical models with efficient numerical schemes for many important applied problems in nondestructive detection, biomedical imaging, dynamic image processing. Under the support of this project, younger resarch teams in the areas of media imaging with international peer-view are educated.
英文关键词: Media imaging;mathematical modeling;ill-posed problems;regularizations;numerics