基于支撑描述的SIFT匹配方法  

SIFT matching method based on support description

作　　者：郑红[1] 刘振强[1] 温天骁 

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100191

出　　处：《北京航空航天大学学报》2014年第5期685-689,共5页Journal of Beijing University of Aeronautics and Astronautics

基　　金：国家自然科学基金资助项目(60543006);博士点基金资助项目(201003259)

摘　　要：为减少局部结构相似等因素导致的图像匹配错误,提出一种基于支持描述的匹配判定方法.利用尺度不变特征变换(SIFT,Scale Invariant Feature Transform)算法获得初始匹配集,选取其中稳定性较高的特征点对建立支撑特征集;根据支撑特征点的分布,对初始匹配集的剩余特征点对进行支撑描述,并根据所生成支撑描述符的相似程度,判定剩余特征点对是否为正确匹配.经判定正确的匹配特征点对被加入支撑特征集,使支撑特征集动态扩展,保证了支撑特征点的分布密度及支撑描述的准确性.实验结果表明,该方法能够在保留正确匹配的同时,消除90%以上的错误匹配,有效提高正确匹配率.To reduce the image matching errors caused by local structure similarity and other factors,a matching judgment method based on support description was proposed. An initial matching set was obtained by scale invariant feature transform（ SIFT） algorithm,from which the more stable feature points were extracted to build a support feature set. According to the distribution of support feature points,a support description on the remaining feature points of the initial matching set was performed. And similarity degree between the generated descriptors was used to determine whether the feature points match correctly. After judgment the correct matching feature points were added to the support feature set,so that the support feature set expanded dynamically and distribution density of the support feature points and accuracy of the support description would be guaranteed. Experimental results show that the proposed method can preserve the correct matches while eliminating more than 90% mismatches and improve the correct matching rate effectively.

关 键 词：图像匹配 支撑描述 尺度不变特征变换 特征描述符 误匹配校正 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]