摘要
针对传统灌溉渠异物入侵监测方法检测精度低、时效性差、夜间巡检不全面、危险性高等问题,提出一种基于雷视融合的灌溉区异物入侵监测方法。针对灌溉区周边行人、动物等小目标误检及特征提取能力不足等问题,提出一种基于YOLOv5改进的小目标识别算法,提高对灌溉区周边小目标检测能力。通过实际场景测试试验,本研究提出的灌溉区雷视融合监测方法和改进的基于YOLOv5的小目标识别算法,识别精确度达到93.26%,监测范围是设备周围360°,有效提升了不同时间段下的异物入侵监测能力,验证了该方法的准确性。
Aiming at the problems of traditional detection of foreign body intrusion in irrigation canals,such as low detection accuracy,poor timeliness and incomplete inspection at night,this study proposed a detection system for foreign body intrusion in irrigation areas based on lightning vision fusion.Meanwhile,aiming at the problems of false detection and insufficient feature extraction ability of small targets such as pedestrians and animals around the irrigation area,this study proposed a small target detection algorithm improved by YOLOv5 to improve the detection ability of small targets around the irrigation area.Through the actual scene test,the proposed detection method of lightning fusion in the irrigation area and the improved small target detection algorithm effectively improved the detection ability of foreign body intrusion and verified the accuracy of the method.
作者
陈晓燕
王川
齐明杰
张宁
林晓龙
霍延强
刘世杰
田源
CHEN Xiaoyan;WANG Chuan;QI Mingjie;ZHANG Ning;LIN Xiaolong;HUO Yanqiang;LIU Shijie;TIAN Yuan(Jinan Water Conservancy Engineering Service Center,Jinan 250013,Shandong,China;Shandong High Speed Construction Management Group Co.,Ltd.,Jinan 250001,Shandong,China;School of Qilu Transportation,Shandong University,Jinan 250002,Shandong,China;School of Microelectronics,Shandong University,Jinan 250101,Shandong,China)
出处
《山东大学学报（工学版）》
CAS
CSCD
北大核心
2024年第3期115-121,共7页
Journal of Shandong University（Engineering Science）
基金
山东省重点研发计划资助项目(2020CXGC010118)。
关键词
激光雷达
摄像机
融合
灌溉区目标检测
小目标检测
LiDAR
the video camera
integration
target detection in irrigation area
small target detection
作者简介
第一作者:陈晓燕(1982-),女,山东梁山人,高级工程师,硕士,主要研究方向为水工结构工程、水利工程管理。E-mail:122473283@qq.com;通信作者:霍延强(2000-),男,山东菏泽人,硕士研究生,主要研究方向为智慧交通车路协同及高速公路差异化收费。E-mail:202215421@mail.sdu.edu.cn。
