华东师范大学学报(自然科学版) ›› 2009, Vol. 2009 ›› Issue (5): 21-28.

• 计算机科学与技术 • 上一篇    下一篇

基于样本块的图像修补技术的改进

唐晓娜， 陈继国， 沈超敏， 张桂戌   

  1. 华东师范大学 计算机科学与技术系, 上海 200241
  • 收稿日期:2008-09-25 修回日期:2008-12-23 出版日期:2009-09-25 发布日期:2014-10-13
  • 通讯作者: 张桂戌

Modified exemplar-based image inpainting algorithm(Chinese)

TANG Xiao-na, CHEN Ji-guo, SHEN Chao-min, ZHANG Gui-xu   

  1. Department of Computer Science and Technology, East China Normal University, Shanghai 200241, China
  • Received:2008-09-25 Revised:2008-12-23 Online:2009-09-25 Published:2014-10-13
  • Contact: ZHANG Gui-xu

PDF

1454

摘要： 改进了一种基于样本块的修补算法，提出了将原本的整体范围搜索改为在用户指定的范围内搜索，最大可能地保证了合适的结构信息被填入，保障修补后的图像在视觉上的连贯性，从而有效提高了大范围破损图像的修补效果.同时，将样本块窗口原有的大小从固定改进为根据图像的实际情况大小可变，使搜索速度得到较大提高.将该算法应用于遥感图像，取得了良好的视觉效果.

关键词: 图像修补, 优先级, 样本块, 搜索匹配, 图像修补, 优先级, 样本块, 搜索匹配

Abstract: This paper changed the searching scope of the original exemplar-based image inpainting from the whole image domain into a user-defined domain, which guarantees that the suitable structure is inpainted, and thereby efficiently improved inpainting results. Meanwhile, the original fixed size of the exemplar was also changed into an automatically adjusted size, which significantly improved the searching speed. This algorithm has been successfully applied to remotely sensed data.

Key words: priority, exemplar, matching, image inpainting, priority, exemplar, matching

中图分类号: