计算机与现代化

• 图像处理 • 上一篇    下一篇

基于结构字典的图像修复算法

  

  1. (北京工业大学计算机学院,北京 100124
  • 收稿日期:2014-04-16 出版日期:2014-07-16 发布日期:2014-07-17
  • 作者简介:刘春荣(1988-),女,山西运城人,北京工业大学计算机学院硕士研究生,研究方向:稀疏表示,图像处理。

Image Inpainting Algorithm Based on Structured Dictionary

  1. (College of Computer Science, Beijing University of Technology, Beijing 100124, China)
  • Received:2014-04-16 Online:2014-07-16 Published:2014-07-17

摘要: 随着稀疏表示理论的日渐完善,利用信号的稀疏性对图像进行修复得到广泛应用。本文针对传统的字典仅是一种无结构的扁平的原子的集合,没有充分利用原子之间相关性的问题,提出基于结构字典的图像修复算法。实验结果表明了该算法的有效性。基于结构字典的图像修复算法不仅可以训练字典更紧致地完成图像修复任务,而且训练得到的字典具有平移不变性、尺度灵活性等优点。

关键词: 结构字典, 图像修复, 稀疏表示, 字典学习

Abstract: With the improvement of sparse representation theory, it is widely used to make full use of sparse property of signals to repair the image. The traditional dictionary has the problem that it cannot make full use of the correlation between dictionary atoms and just a collection of unstructured atoms. Therefore we propose an image inpainting algorithm based on structured dictionary and illustrate the validity in image inpainting tasks. The algorithm can not only finish image inpainting task by training tighten dictionary, but also the dictionary has the merits of shift-invariance and scale flexibility and so on.

Key words: structured dictionary, image inpainting, sparse representation, dictionary learning

中图分类号: