摘要
近年来,具有尺度、旋转不变性的SIFT(scale-invariant feature transform)及其改进算法,在图像匹配中得到了广泛应用并极大提高了匹配精度,但是无论采用何种匹配方法,总会产生误匹配点。因此,误匹配点检测也是图像匹配的一个重要步骤。为了使国内外同行对现有误匹配点检测方法有较全面的了解,对这些方法进行了综述。在参考国内外大量文献的基础上,将现有误匹配点检测方法分为三类:基于函数拟合的方法;基于统计模型的方法;基于图的方法。对以上三类方法进行了综述,并对各个方法的优缺点进行了分析;最后,提出了误匹配点检测中还需要解决的一些问题,并给出了基本的解决思路。
Scale-invariant feature transform（SIFT） and some improvements are invariant to image scaling and rotation and have been widely used in image matching. These methods have significantly improved image matching accuracy. However, false matching points still occur in all image matching algorithms. Therefore, false matching points detection is an important step in image matching. In order to give a comprehensive understanding of false matching points detection methods to researchers, this paper reviewed false matching points detection methods. It categorized there methods into three classes based on numerous researches：the method based on function fitting, the method based on statistical model, the method based on graph. Then,it re- viewed these three methods and analyzed the advantages, disadvantages and suitable applying conditions of these methods. Fi- nally, it proposed some problems needed to solve in the future and gave some solutions for these problems.
出处
《计算机应用研究》
CSCD
北大核心
2015年第9期2561-2565,2571,共6页
Application Research of Computers
基金
国家"863"计划资助项目(2012AA12A304
2013AA12A301)
