Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (26): 191-193.DOI: 10.3778/j.issn.1002-8331.2010.26.059

• 图形、图像、模式识别 • Previous Articles     Next Articles

Structure of local feature for target recognition

CHENG Lei,WU Xiu-qing   

  1. Department of Electronic Engineering and Information Science,University of Science and Technology of China,Hefei 230027,China
  • Received:2009-02-18 Revised:2009-04-13 Online:2010-09-11 Published:2010-09-11
  • Contact: CHENG Lei

局部特征几何结构用于目标识别

程 蕾,吴秀清   

  1. 中国科学技术大学 电子工程与信息科学系,合肥 230027

  • 通讯作者: 程 蕾

Abstract: Recently local feature based target recognition is becoming one of hot topics in computer vision.On the base of bag-of-word method only using appearance,this paper proposes a star model,which connects the location information of local key-point,composed of some parts depending on a landmark.It introduces the structure of image local region and reduces the complexity of full model.Experiments show that the proposed algorithm outperforms bag-of-word method in classification and realizes localization effectively.

Key words: local feature, structure, target recognition

摘要: 基于局部特征的目标识别成为计算机视觉研究中的一个热点。该文在词袋方法仅利用目标外观信息的基础上,将局部特征点的位置信息建模为星形模型,模型由固定部分组成且仅依赖于标志位置。引入图像局部区域的几何结构信息,同时降低了全模型中各个部分依赖的复杂性。实验证明该方法的分类效果优于词袋方法,并能有效实现目标定位。

关键词: 局部特征, 几何结构, 目标识别

CLC Number: