基于改进ORB算法的特征点匹配  

Feature Point Matching Based on Improved ORB Algorithm

作　　者：胡志锋 许钢[1] 陈玲 伏娜娜 HU Zhifeng;XU Gang;CHEN Ling;FU Na′na(Anhui Key Laboratory of Testing Technology and Energy Saving Equipment, Anhui University of Engineering, Wuhu, Anhui 241000,China)

机构地区：[1]安徽工程大学检测技术与节能装置安徽省重点实验室,安徽芜湖241000

出　　处：《平顶山学院学报》2022年第2期55-60,共6页Journal of Pingdingshan University

基　　金：安徽省高校自然科学研究重点项目(KJ2018A0111);安徽省重点实验室开放项目(2017070503B026-A01)。

摘　　要：针对传统ORB算法不具有尺度不变性的问题及匹配速率的问题,提出改进ORB算法的特征匹配.将SURF算法与传统ORB算法结合,先利用SURF算法的尺度金字塔得到具有尺度不变性的特征点,解决特征点匹配的尺度性问题,再对用ORB算法生成的高维描述符进行降维处理,提高算法的匹配速率,最后用暴力匹配方式完成图像匹配.实验结果解决了传统ORB算法在特征匹配时的尺度不变性问题,同时降维处理提高了算法匹配速率.Aiming at the problem that the traditional ORB algorithm in visual odometer does not have scale invariance and the speed of matching,the improved feature matching of ORB algorithm is proposed.Combining SURF algorithm with traditional ORB algorithm,the feature points with scale invariance are obtained by using the scale pyramid of SURF algorithm to solve the scale problem of feature point matching,Then,the dimension reduction processing is carried out for the high-dimensional descriptor generated by ORB algorithm to improve the matching rate of the algorithm.Finally,the image matching is completed by violent matching.The experimental results demonstrate that the proposed method could solve the problem of scale invariance of traditional ORB algorithm in feature matching.At the same time,the dimension reduction processing improves the matching rate of the algorithm.

关 键 词：ORB算法 尺度不变性 降维处理 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]