计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (29): 230-232.DOI: 10.3778/j.issn.1002-8331.2009.29.069

• 工程与应用 • 上一篇    下一篇

基于快速连通域分析的目标特征提取算法

张 恒,胡文龙,丁赤飙   

  1. 中国科学院 电子学研究所,北京 100190
  • 收稿日期:2008-06-03 修回日期:2008-09-22 出版日期:2009-10-11 发布日期:2009-10-11
  • 通讯作者: 张 恒

Fast algorithm for target feature extraction based on connected component analysis

ZHANG Heng,HU Wen-long,DING Chi-biao   

  1. Institute of Electronics,Chinese Academy of Sciences,Beijing 100190,China
  • Received:2008-06-03 Revised:2008-09-22 Online:2009-10-11 Published:2009-10-11
  • Contact: ZHANG Heng

摘要: 基于子区域思想,提出了一种快速的连通域分析方法用于提取视频目标特征。算法只需对图像进行一遍扫描,就可以用一棵多叉树描述出连通域各子区域之间的等价关系,同时由多叉树支节点向根节点完成目标特征提取并将特征保存在多叉树的根节点中。算法占用内存小,具有很高的运行效率。

关键词: 行程, 子区域, 特征提取, 连通域分析

Abstract: Based on the idea of sub-region,a fast algorithm for Connected Component Analysis(CCA) is proposed to extract features of video targets.Through scanning an image for only one time,the algorithm can describe the equivalences between sub-regions belonging to the same connected component with a multi-tree.And at the same time,features of a target are extracted from leaf nodes to the root node and stored in the root node.It can be demonstrated that the proposed algorithm shows low space and high efficiency by analysis and experimentation.

Key words: run-length, sub-region, feature extraction, connected component analysis

中图分类号: