计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (22): 195-196.DOI: 10.3778/j.issn.1002-8331.2008.22.058

• 图形、图像、模式识别 • 上一篇    下一篇

基于修复顺序的图像修复算法

张伟彬   

  1. 信息产业部 电信科学技术第一研究所,上海 200032
  • 收稿日期:2007-10-11 修回日期:2008-01-16 出版日期:2008-07-11 发布日期:2008-07-11
  • 通讯作者: 张伟彬

Image repair based on repairing sequence

ZHANG Wei-bin   

  1. The First Research Institute of Telecom of Ministry of Information Industry,Shanghai 200032,China
  • Received:2007-10-11 Revised:2008-01-16 Online:2008-07-11 Published:2008-07-11
  • Contact: ZHANG Wei-bin

摘要: 提出一个用于图像修复的算法。首先计算修复前沿每个待修复小片的优先级,然后优先修复具有最高优先级的小片。优先级计算方法能体现每个小片具有的结构信息以及已有颜色的可信度,使得基于小片的纹理综合方法有序进行,同时修复目标区域中的纹理和结构信息。用户选择一个目标区域后,算法就根据选择区域周围的信息自动地修复目标区域。实验证明该算法不仅快速且修复的效果非常好,且目标区域较大时也能达到很好效果。

关键词: 图像修复, 纹理综合, 纹理, 结构

Abstract: This paper proposes an algorithm for image repair.This algorithm performs the task through a best-first repairing strategy based on the priority assigned to each patch on the repairing front.The priority computation is biased toward those patches with structure information and surrounded by high-confidence pixels,so that the texture synthesis is performed orderly.This algorithm can not only synthesize texture but also structure information.After the object area has been selected by the user,this algorithm will perform repairing totally automatically.Experiment results show that this algorithm is fast and perfect even if the object area is big.

Key words: image repair, texture synthesis, texture, structure