摘要
提出一种单目测距算法,将车牌检测与距离测量分离,以车牌作为车牌检测和距离测量的纽带,使用YOLOv4-Tiny pro检测车牌,基于几何关系的车辆测距模型,根据车牌尺寸的先验信息准确计算出车辆距离。最后对系统的识别效果和距离检测结果分别进行了实验验证,对于整合后的算法进行了进一步优化,证明了方法的有效性。
The license plate detection was separated from the distance measurement,and the license plate was used as the link between the license plate detection and the distance measurement.YOLOv4-Tiny pro was used to detect the license plate,and the vehicle distance was calculated accurately based on the prior information of the license plate size.Finally,the recognition effect and distance detection results of the system are verified by experiments,and the integrated algorithm is further optimized to prove the effectiveness of the method.
作者
张凯龙
钱莉
陈梦涛
陆洪光
单明陶
杜嘉伟
Zhang Kailong;Qian Li;Chen Mengtao;Lu Hongguang;Shan Mingtao;Du Jiawei(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201600,China)
出处
《农业装备与车辆工程》
2022年第11期53-57,共5页
Agricultural Equipment & Vehicle Engineering
基金
2020年度上海工程技术大学研究生科研创新项目(20KY0111)。
关键词
车辆距离
深度学习
单目测距模型
相似三角形
vehicle distance
deep learning
monocular ranging model
similar triangle