项目名称: 基于一类线性增长非凸泛函的图像恢复研究
项目编号: No.11301113
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 数理科学和化学
项目作者: 郭志昌
作者单位: 哈尔滨工业大学
项目金额: 23万元
中文摘要: 图像恢复问题是典型的求解反问题的过程,其本身求解的病态特征,决定了图像恢复工作并不是一件简单的事情。本项目旨在研究一类线性增长非凸泛函在图像恢复中的数学问题和应用研究,主要内容包括:1)提出线性增长图像恢复框架,分别讨论当泛函满足严格凸、拟凸以及非凸不同限制条件下能量泛函的性质以及在图像处理中的反应;2)在BV空间里考虑线性增长泛函解的适定性,以及对应演化反应扩散方程解的存在唯一性和渐近性;3)讨论泛函凸性与正倒向扩散的关系,提出可控正倒向扩散图像恢复模型,并讨论方程解的适定性与渐进性;4)提出合理的数值格式,并且进行参数分析,实现参数的自适应选取。通过研究,得到高效的、高信噪比的图像恢复滤波器;发展变分法在图像恢复模型中的应用;完善非线性扩散方程的理论。本项目是对多学科交叉融合的促进,它不仅具有很大的实用价值,而且具有重要的理论价值和科学意义。
中文关键词: 图像恢复;线性增长泛函;BV空间;正倒向扩散;非凸
英文摘要: The image restoration is a type of inverse problems and the problems are ill-posed, which make it not easy to deal with. This project considers a class of linear growth nonconvex functionals in image restoration, the main contents includes: 1)Proposing the linear growth functional framework for image restoration, we consider the properties of the energy functionals and the associated reaction-diffusion equations for the image restoration under the different cases, such as the strictly convex quasi-convex and nonconvex, respectively;2)We consider the posedness of the solutions of linear growth functional in the BV space and the existence and uniqueness and asymptotic behavior of reaction diffusion equations;3) Inspired the nonconvex functional, we proposed the controlled forward-backward diffusion equations and consider the posedness and asymptotic behavior of the equations; 4) For new models, we propose the new numerical schemes and anlysis the paremeters to implement the adaptive selection of them. Through the research, the purpose of this project is to obtain an efficient and high SNR image restoration filter; develop the variational method on the application of image restoration; perfect the theory of the nonlinear diffusion equation; obtain efficient numerical schemes. The project is to promote the combi
英文关键词: Image restoration;Linear growth functional;BV-space;Forward-backward diffusion;Nonconvex