作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (16): 232-234. doi: 10.3969/j.issn.1000-3428.2011.16.079

• 图形图像处理 • 上一篇    下一篇

图像修补的优化变换方法

聂笃宪 1,李 杰 2,陈鹤峰 3   

  1. (1. 华南农业大学理学院,广州 510642;2. 湘潭大学土木工程与力学学院,湖南 湘潭 411105;3. 广东工业大学数学学院,广州 510090)
  • 收稿日期:2011-02-21 出版日期:2011-08-20 发布日期:2011-08-20
  • 作者简介:聂笃宪(1974-),男,讲师、博士研究生,主研方向:图像处理,最优化方法;李 杰,讲师、硕士;陈鹤峰,讲师、博士研究生
  • 基金资助:
    国家自然科学基金资助项目(50775079);国家青年基金 资助项目(60702030);华南农业大学校长基金资助项目(2009K023)

Optimization Transformation Method for Image Inpainting

NIE Du-xian 1, LI Jie 2, CHEN He-feng 3   

  1. (1. College of Science, South China Agricultural University, Guangzhou 510642, China; 2. College of Civil Engineering and Mechanics, Xiangtan University, Xiangtan 411105, China; 3. College of Mathematics, Guangdong University of Technology, Guangzhou 510090, China)
  • Received:2011-02-21 Online:2011-08-20 Published:2011-08-20

摘要: 采用整体变分(TV)模型修补图像,提出一种图像修补的优化变换方法。引入一个辅助变量,利用优化变换,将TV模型中单变量函数的优化问题转化为等效双变量函数的优化问题,并利用交替迭代最小化算法和Chambolle’s投影算法求解模型。实验结果表明,与采用梯度下降法的TV模型算法相比,该方法的图像修补效率和修补效果较优。

关键词: 整体变分, 图像修补, 优化变换, 交替迭代最小化算法

Abstract: This paper applies Total Variational(TV) model to inpaint image, proposes an optimization transformation method for image inpainting. It uses an optimization transformation technique which involves replacing their univariate functional of TV model by a bivariate functional by adding an auxiliary variable. The Chambolle’s projection algorithm and alternating minimization method are employed in solving the model. Experimental results show that compared with the algorithm of solving TV model based on gradient descent method, the efficiency and quality of inpainted images by the proposed method is improved.

Key words: Total Variation(TV), image inpainting, optimization transformation, alternating minimization algorithm

中图分类号: