摘要
针对传统图形匹配算法对稳定特征提取不充分的缺点,提出了一种基于MSER与SURF的图形匹配新方法。MSER与SURF是两种特征提取算法,各有优缺点,且具有互补的特性。提出的算法分别用MSER与SURF检测图像的特征点,用SURF描述子表征检测到的所有特征点,从而实现了两者的互补,并获取了更为丰富的特征描述。基于更丰富的特征,描述信息,进行特征匹配,最后得到的图像匹配效果,相比传统方法更加稳定。
Aimed at the defect of traditional image registration method, which detect the stable features inadequately, this paper proposes a new image registration method based on MSER and SURF.MSER and SURF are two different feature extraction methods, and they have complementary advantages.The presented method uses MSER and SURF to detect the features, SURF descriptor to represent all extracted features by MSER and SURF, resulting better features than both MSER and SURF. Based on better feature representation to match the extracted features, finally the image registration obtained is much robust than traditional method.
出处
《微型电脑应用》
2012年第3期61-64,69,共4页
Microcomputer Applications