摘要
针对光流法定位精度差的问题,设计了一种光流和改进定向二进制简单描述符(ORB)算法融合的单目视觉里程计方法。首先利用光流法进行帧间位移定位,并根据经验设定每帧跟踪的最少特征点数,在跟踪特征点数低于设定的阈值后,利用改进的ORB算法进行帧间位移定位,最后通过二者的循环运行更新机器人的位置和姿态。结果表明:该方法克服了光流法定位精度差的缺点,突出了改进ORB算法定位准确性的优点,能够提供较准确的定位输出。
Aiming at the problem of poor positioning precision of optical flow method,a monocular vision odometer method based on fusion of optical flow and improved orientated binary(ORB) simple descriptor algorithm is designed. Firstly,the optical flow method is used to locate the inter-frame displacement,and the minimum number of feature points per frame is set according to the experience. After the number of tracking feature points are lower than the set threshold,the improved ORB algorithm is used to locate the inter-frame displacement. Finally,the position and attitude of the robot are updated through the cyclic operation of the two. The results show that this method overcomes the shortcomings of poor positioning precision of optical flow method,highlights the advantages of positioning accuracy of improved ORB algorithm,and can provide more accurate positioning output.
作者
唐浩
王红旗
茹淑慧
TANG Hao;WANG Hongqi;RU Shuhui(School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo 454000,China)
出处
《传感器与微系统》
CSCD
2019年第12期83-85,共3页
Transducer and Microsystem Technologies
基金
河南省科技攻关计划项目(172102210270)
关键词
单目视觉里程计
光流
改进ORB算法
阈值
monocular vision mileage
optical flow
improved ORB algorithm
threshold
