Heat flows and related minimization problem in image restoration

被引:23
作者
Barcelos, CAZ
Chen, Y
机构
[1] Univ Fed Uberlandia, Dept Matemat, BR-38400 Uberlandia, MG, Brazil
[2] Univ Florida, Dept Math, Gainesville, FL 32611 USA
基金
美国国家科学基金会;
关键词
image restoration; total variation; heat flow method; anisotropic diffusion;
D O I
10.1016/S0898-1221(00)00048-1
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
A new anisotropic diffusion model is proposed for image restoration and segmentation, which is closely related to the minimization problems for the unconstrained total variation E(u) = integral(Omega) alpha(x)\del u\ + (beta/2)\u - I\(2). Existence, uniqueness, and stability of the viscosity solutions of the equation are proved. The experimental results are given and compared with the existing models in the framework of image restoration. The improvement on preserving sharp edges by using the new model is visible. (C) 2000 Elsevier Science Ltd. All rights reserved.
引用
收藏
页码:81 / 97
页数:17
相关论文
共 12 条
[11]  
Strong D.M., 1997, THESIS U CALIFORNIA
[12]   Iterative methods for total variation denoising [J].
Vogel, CR ;
Oman, ME .
SIAM JOURNAL ON SCIENTIFIC COMPUTING, 1996, 17 (01) :227-238