摘要
为解决塔吊作业中人员进入吊装危险区域难以及时有效识别的问题,提出一种塔吊多维动态空间融合监控视频流数据下的危险区域人员入侵智能监测方法。首先构建BIM模型+视频流的底层数据,将塔吊作业区BIM多维空间模型与高视角监控视频流数据进行融合;然后接入开源的YOLOv5目标检测与定位算法,持续获取吊钩和人员在监控摄像机视场内的像素坐标序列,以吊钩垂直投影点为基准结合视频标定设立虚拟多维塔吊动态危险区域电子围栏,运用坐标转换技术将获取的人员轨迹坐标和围栏坐标统一转化为三维局部世界坐标;最后将以上数据同步映射于同一BIM模型中,通过判别算法实现人员入侵动态危险区域的实时有效判断。实例结果表明,该方法有效、可行,可以为塔吊作业的安全管理提供新的视角。
To address the challenge of timely identification for personnel intrusion into hazardous areas during tower crane operations,this study proposes an intelligent monitoring method integrating multi-dimensional dynamic spatial data with video streams.First,a foundational dataset combining BIM models and surveillance video streams is established by fusing a BIM-based multi-dimensional spatial model of crane operation zones with high-angle monitoring video streams.Then,the open-source YOLOv5 algorithm is employed for continuous detection and positioning,acquiring pixel coordinate sequences of hooks and personnel within the camera's field of view.A virtual electronic fence for dynamic hazardous zones is created based on the hook's vertical projection and video calibration,while coordinate conversion technology transforms personnel trajectories and fence boundaries into a unified 3D local-world coordinate system.Finally,synchronized mapping of these data onto the BIM model enables real-time intrusion detection via a discriminant algorithm.Experimental results demonstrate the method's effectiveness and feasibility,offering new insights for safety management in crane operations.
作者
赵平
刘飞
刘倩
刘钊
ZHAO Ping;LIU Fei;LIU Qian;LIU Zhao(School of Resources Engineering,Xi'an University of Architecture and Technology,Xi'an Shaanxi 710055,China;School of Civil Engineering,Xi'an University of Architecture and Technology,Xi'an Shaanxi 710055,China)
出处
《工业安全与环保》
2025年第6期66-69,共4页
Industrial Safety and Environmental Protection
关键词
视频融合
多维动态空间
YOLOv5
位置映射
入侵监测
video fusion
multidimensional dynamic space
YOLOv5
location mapping
intrusion detection
