摘要
快速、正确的匹配从两幅或多幅图像中提取出来的特征点是基于特征点图像配准问题的关键。传统的只使用归一化互相关匹配(NCC)算法进行的特征点粗匹配,虽然具有较强的抗噪声能力,但是匹配的速度很慢,而且错误率也比较高。因此,在研究了NCC算法与序贯相似度检测(SSDA)算法基础上,并改进了SSDA算法的非相似度计算方法,提出将两种算法融合在一起形成一种快速的特征点匹配算法,改进算法充分利用了两种算法的优点,大大提高了特征点的匹配速度,而且减少了错误匹配的个数。通过实验证明,该算法是一种有效的特征点匹配算法,比只使用NCC算法进行特征点匹配所需的时间降低了70%以上,正确匹配率也有所提高。
Fast and correct matching feature points which extracted from two or more images is the key problem to base on feature points of image registration.The traditional use only normalized cross-correlation matching（NCC） algorithm for coarse matching feature points,although with a strong anti-noise capability,but the match is very slow and the error rate is relatively high.After researching the NCC and SSDA algorithm,and improving the method of non-similarity of SSDA,the paper fuses the two algorithms to form a fast feature point matching algorithm.The algorithm makes full use of advantages of the two algorithms;it highly improves the speed of feature point＇s match,and reduces the amount of wrong match.Experiments show that this method is an effective algorithm of feature point match.It improves more then 70% of feature point＇s match and improves correct match rate.
出处
《计算机与数字工程》
2010年第10期19-21,64,共4页
Computer & Digital Engineering
关键词
HARRIS
归一化互相关
序贯相似度检测算法
角点匹配
图像配准
Harris
normalized cross correlation（NCC）
sequential similarity detection algorithm（SSDA）
corner matching
image registration
