摘要
为提高学生竞赛系统中机器人定位和目标检测的准确率,提出基于ViBe算法的ORB特征匹配方法。首先,对ORB算法原理进行具体分析;然后加入四叉树分块用于前景分割,以减少图像边缘和纹理较少的目标点,之后再进行特征匹配,减少提取特征点数量,并简化算法过程;然后在改进的ORB特征匹配方法基础上,加入ViBe运动目标检测算法对ORB竞赛系统机器人进行更精准的特征匹配;最后将改进算法应用到学生竞赛系统机器人中进行实验验证。实验结果表明,提出的改进ViBe算法的精确率和误检率分别为78.4%和6.5%,均优于传统深度学习算法,由此说明基于改进的ViBe算法的特征匹配率更高,鬼影消除效果较好,匹配速度显著提升,可实现学生竞赛系统中机器人的精准定位和目标检测。
To improve the accuracy of robot positioning and target detection in the student competition system,an ORB feature matching method based on the ViBe algorithm is proposed.Firstly,the ORB algorithm principle;then quadruple tree is added for foreground segmentation to reduce image edges and textures,then feature matching,reduce the number of extraction points and simplify the algorithm process;Based on the improved ORB feature matching method,ViBe motion target detection algorithm for more accurate feature matching of the ORB competition system robot;Finally,the improved algorithm is applied to the student competition system robot for experimental verification.The experimental results show that the accuracy rate and error detection rate of the proposed improved ViBe algorithm are 78.4%and 6.5%,respectively,which are better than the deep learning algorithm,which shows that the feature matching rate of the improved ViBe algorithm is higher,the ghost elimination effect is better,and the matching speed is significantly improved,which can realize the accurate positioning and target detection of robots in the student competition system.
作者
赵瑞丹
冯雨
ZHAO Ruidan;FENG Yu(Xi'an Aeronautical Polytechnic Institute,Xi'an 710089,China)
出处
《自动化与仪器仪表》
2022年第3期172-176,共5页
Automation & Instrumentation
基金
陕西省教育改革研究项目:互联网+SPOCS背景下在线开放课程的改革与实践(17GY009)。
关键词
ORB特征匹配
ViBe算法
四叉树分块
目标检测
ORB feature matching
ViBe algorithm
quadruple tree segmentation
target detection