摘要
对于在传统的图像匹配过程中,存在误匹配率高和匹配效果不佳的问题,提出基于加速鲁棒特征(SURF)算法与快速近似最近邻查找(FLANN)搜索的图像匹配方法。首先采用Hessian矩阵来获知图像的局部最值,然后在图像上构建尺度空间,通过不同的尺度空间定位出特征点,并确立特征点的主方向,再生成特征点描述子,最后结合FLANN搜索算法对图像进行匹配。实验表明,该算法相对传统的图像匹配方法提高准确度和匹配效果。
For the problem of high mismatching rate and poor matching effect in the traditional image matching process,a new image matching meth⁃od based on Speed Up Robust Feature(SURF)algorithm and Fast Library for Approximate Nearest Neighbors(FLANN)search is pro⁃posed.First,Hessian matrix is used to obtain the local maximum value of the image.Then,scale space is constructed on the image.Feature points are located through different scale Spaces.Experimental results show that the algorithm is more accurate and efficient than the tradi⁃tional image matching method.
作者
徐明
刁燕
XU Ming;DIAO Yan(College of Mechanical Engineering,Sichuan University,Chengdu 610065)
出处
《现代计算机》
2020年第14期49-52,57,共5页
Modern Computer
基金
四川大学泸州市人民政府战略合作项目(No.2018CDLZ-22)。
关键词
SURF
特征提取
FLANN
图像匹配
SURF(Speed Up Robust Feature)
Feature Extraction
FLANN(Fast Library for Approximate Nearest Neighbors)
Image Matching
