摘要
尺度不变的特征变换方法(SIFT)具有对图像尺度缩放、旋转、放射变换以及亮度变化保存不变的优点,该文提出了一种基于SIFT算法的精准的图像配准方法。首先利用SIFT算法提取图像中的特征点;接着对这些特征点采用相似性准则中的欧式距离进行匹配,得到初始匹配对;由于初始匹配对中存在误匹配对,从而造成匹配的精度不足,因此提出一种改进后的RANSAC(随机取样一致性性算法)消除误匹配对。此外该文通过多次实验,选取SIFT算法中合适的比例阀值,提高配准的精度。实验结果表明,该方法既继承了SIFT算法的鲁棒性,又提高了匹配精度。
Scale Invariant Feature transform(SIFT)has advantages of scale scaling,rotating,radiation transform and unchanged brightness changes on image.A kind of precise SIFT algorithm based on image registration is proposed.First,Extract feature points of images using the SIFT algorithm.Then,get initial matching points from these feature points by the matching method of similarity criterion in Euclidean distance.The improved RANSAC which for eliminating mismatch points is proposed due to existing mismatching points in initial matching points,causing the matching of precision insufficient.Furthermore,in order to improve the accuracy,we select suitable threshold in SIFT through multiple tests.The experiments shoe that this method both inherit the robustness of SIFT and improve the precision of matching.
出处
《电脑知识与技术（过刊）》
2011年第1X期400-402,共3页
Computer Knowledge and Technology
