北京航空航天大学学报 ›› 2008, Vol. 34 ›› Issue (05): 516-519.

• 论文 • 上一篇    下一篇

一种基于图像特征点提取及匹配的方法

张少辉, 沈晓蓉, 范耀祖   

  1. 北京航空航天大学 自动化科学与电气工程学院, 北京 100083
  • 收稿日期:2007-05-08 出版日期:2008-05-31 发布日期:2010-09-17
  • 作者简介:张少辉(1982-),男,山东青岛人,硕士生,zhanahohui-3@163.com.
  • 基金资助:

    北京市教育委员会共建项目(BHBJZD-1-5)

Method in image-s feature extraction and matching

Zhang Shaohui, Shen Xiaorong, Fan Yaozu   

  1. School of Automation Science and Electrical Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
  • Received:2007-05-08 Online:2008-05-31 Published:2010-09-17

摘要: 针对图像特征提取与匹配的适应性和准确性的问题,将尺度不变特征变换(SIFT, Scale Invariant Feature Transform)算法应用到图像匹配领域.首先从原理上对SIFT算法的特性进行了分析,并以visual studio 2005为开发平台对SIFT算法分步骤进行了实现;最后以基于欧氏距离的最近邻准则作为特征的相似度量将SIFT算法提取的特征应用于图像特征匹配,并对不同的近邻比进行比较,给出了建议值.通过3组实验图像的匹配结果表明,SIFT算法提取的特征对图像缩放、旋转、亮度变化的匹配正确率都等于或接近100%,证明了SIFT算法提取的特征点有很好的适应性和准确性,可以进一步应用到图像识别以及图像重建等领域.

Abstract: To solve the problem of adaptability and accuracy in the field of image feature extraction and feature matching, the method of scale invariant feature transform (SIFT) was introduced. Firstly the characteristics of the SIFT method were analyzed by theory, and the SIFT method was implemented step by step on the visual studio 2005 platform; Then the features extracted by SIFT method were applied to match images on the criterion of nearest neighbor based on Euclidean distance. A suggestion value bound was given by comparing the matching result of different nearest ratio. At last the effect of the SIFT method was validated by the matching result of three different groups of images. The matching result shows that the features extracted by SIFT method are invariant to image scale, rotation and illumination change, and the matching accuracies are all equal or close to 100%. These results prove that the features extracted by SIFT method have excellent adaptive and accurate characteristics, which are useful for the fields of image recognition, image reconstruction, etc.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发