【作者】 艾浩军； 朱荣； 张敏； 李俊； 方禹； 王红霞；

【Author】 Ai Haojun Zhu Rong Zhang Min Li Jun Fang Yur 1, 2 , Wang Hongxia 2 1 School of Computer Science, Wuhan University, Wuhan 430072, China 2 National Engineering Research Center for Multimedia Software, Wuhan 430072, China 3 Department of Computer Science and Technology, Wuhan University of Technology, Wuhan 430063, China

【机构】 武汉大学计算机学院； 国家多媒体软件工程技术研究中心； 武汉理工大学计算机科学与技术学院；

【摘要】 SIFT描述子是物体识别中应用最广泛的局部描述算子之一,分析了SIFT关键点匹配的尺度空间相关性,提出了一种尺度分量用于SIFT关键点的匹配算法,以此为基础,设计了由粗到精两级匹配算法。实验表明,与最近邻准则和比值准则相比,综合查全率和精度的指标有明显改善。更多还原

【Abstract】 SIFT descriptor was used widely in object recognition as a local description feature. We analyzed the correlation between the distribution of the matched SIFT key points and the scale space. We proposed a two-stage matching algorithm, in which the scale component of SIFT was used in the second stage. This module was embedded into a hierarchical clustering algorithm. The experiments show that it is better than the nearest neighbor criteria and the ratio threshold criteria by comparing the recall and the precision.更多还原