一种基于特征点的图像匹配新方法  

New Method of Image Matching Based on Characteristic Points

作　　者：李为华[1] 苏辉 

机构地区：[1]信阳师范学院计算机与信息技术学院,河南信阳464000 [2]信阳师范学院网络信息与计算中心,河南信阳464000

出　　处：《湖北第二师范学院学报》2017年第8期25-28,共4页Journal of Hubei University of Education

基　　金：河南省高等学校重点科研项目(17B520034);河南省教育厅教改研究项目(2017-JSJYYB-055)

摘　　要：如何从浩如烟海的图像数据库中迅速并准确地进行图像匹配成为数字生活中的一项难题。为了解决这个难题,本文提出一种图像匹配新方法:针对FAST算法噪声敏感不足、不具有尺度不变性进行改进——得到Zoser-FAST算法;结合SIFT算法,以描述符间的点积作为相似性度量进行图像匹配。仿真实验结果表明改进的图像匹配方法大大减少了运算量,加快了图像匹配的运算速度、提高了准确率。How multimedia could quickly and accurately match an image from the vast image database has become a problem in the digital life. In order to solve the problem,a new method of image matching is proposed in this paper： from FAST algorithm because it has inadequate noise sensitiveness and has no scale invariance,Zoser-FAST algorithm is developed; image matching is carried out with descriptor＇s scalar product as similarity measurement. According to the simulation experiment,the improved image matching method could greatly decrease the work load of calculation,speed up the calculation of image matching and promote the accuracy.

关 键 词：图像匹配 SIFT 尺度空间 Zoser-FAST 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]