Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (24): 215-218.DOI: 10.3778/j.issn.1002-8331.2008.24.065

• 工程与应用 • Previous Articles     Next Articles

Approach for natural landmark matching and tracking used in robot navigation

LIU Jun-cheng,WANG Miao-xin   

  1. Department of Automation,North China Electric Power University,Beijing 102206,China
  • Received:2008-01-28 Revised:2008-03-07 Online:2008-08-21 Published:2008-08-21
  • Contact: LIU Jun-cheng

一种机器人导航中自然路标的匹配与跟踪方法

刘俊承,王淼鑫   

  1. 华北电力大学 自动化系,北京 102206
  • 通讯作者: 刘俊承

Abstract: This paper describes an approach for natural landmark matching and tracking based on SIFT and KLT methods.In this approach,uses SIFT algorithm to extract the key points set from selected landmark as model points set,then the affine transformation of image key points to model key points set will be acquired.Mobile robot can use this information for self-localization.When natural landmark found,the KLT algorithm will be used to track the matched landmark.The experiments show the efficiency and robustness of the proposed approach in natural landmark matching and tracking.

Key words: natural landmark, Scale Invariant Feature Transform(SIFT), Kanade-Lucas-Tomasi(KLT), key points matching, key points tracking

摘要: 提出了一种基于SIFT和KLT算法的自然路标匹配与跟踪方法。该方法利用SIFT算子提取图像中自然路标的特征点集作为模板,然后将机器人采集图像中的SIFT特征点集与模板特征点集进行匹配,获取二者之间仿射关系,并解算自然路标在视野中的位置,为机器人自定位提供参考信息。机器人在运行过程中,将KLT算法与SIFT算法相结合对成功匹配的自然路标进行跟踪,较好地解决了SIFT算法效率低下的问题。实验结果表明该方法对自然路标具有较好的匹配和跟踪效果。

关键词: 自然路标, SIFT, KLT算法, 特征点匹配, 特征点跟踪