计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (22): 56-58.

• 学术探讨 • 上一篇    下一篇

基于统计特征的图像修复

陈 芳1,叶正麟1,王继红2   

  1. 1.西北工业大学 理学院,西安 710072
    2.西北工业大学 软件与微电子学院,西安 710072
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-08-01 发布日期:2007-08-01
  • 通讯作者: 陈 芳

Image inpainting based on statistical features

CHEN Fang1,YE Zheng-lin1,WANG Ji-hong2   

  1. 1.School of Science,Northwestern Polytechnical University,Xi’an 710072,China
    2.Software and Microelectronics College,Northwestern Polytechnical University,Xi’an 710072,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-08-01 Published:2007-08-01
  • Contact: CHEN Fang

摘要: 在基于样图图像修复思想的基础上,从填充顺序和匹配准则两个方面进行改进,提出了一种图像修复方法。优先级函数加入了面片梯度信息和方差信息,使得填充顺序能够正确地修复结构和纹理信息。匹配准则采用自定义的颜色特征函数和统计特征函数对相似片进行两重匹配,取最优匹配块填充。一系列的修复实例证明该方法的有效性和鲁棒性。

关键词: 图像修复, 纹理合成, 颜色特征函数, 统计特征函数

Abstract: An image inpainting method based on the existing exemplar-based image inpainting idea is proposed.Improves the method by the filling order and matching criterion.The priority is added in gradient information and variance information.It makes sure that the filling order can propagate texture and structure information correctly.The matching criterion is added in the color eigenfunction and statistical eigenfunction.Proposes a two-pass matching criterion to get the optimum matching patch.A number of examples demonstrate the effectiveness and the robustness of our algorithm.

Key words: image inpainting, texture synthesis, color eigenfunction, statistical eigenfunction