摘要
电气开关指示位置快速识别是MR智能交互系统的基础服务。为了尽量减少计算负担,首先分析了不同类型开关的状态特征,介绍了MR智能眼镜在开关指示位置识别时的优势,然后根据重要程度不同,应用激光点云扫描技术生成兼顾精度和数据量的设备对象及周边环境三维模型,最后采用基于特征点法的视觉SLAM算法和YOLOv3算法技术完成开关指示位置的定位与快速识别。实际应用表明,对于设备上的开关指示位置,基于MR技术的电气开关指示位置快速识别方法的识别率高达100%,满足MR智能交互系统的需求。
The rapid identification of the indicated position of electrical switches is the basic service of the MR intelligent interactive system.To minimize the computational burden,the state characteristics of different types of switches are analyzed,and the advantages of MR smart glasses in switch indication position recognition are introduced.Then,according to the different importance,the laser point cloud scanning technology is used to generate a three-dimensional model of the equipment object and the surrounding environment that considers both accuracy and data volume.Finally,visual SLAM algorithm based on feature point method and YOLOv3 algorithm technology are used to locate and quickly identify the switch indication position.The practical application shows that for the switch indication position on the equipment,the recognition rate of the quick identification method of electrical switch indication position based on MR technology is up to 100%,which meets the requirements of MR intelligent interactive system.
作者
方陈涛
王贤湖
胡科
FANG Chentao;WANG Xianhu;HU Ke(Fengxian Power Supply Company,State Grid Shanghai Electric Power Company,Shanghai 201499,China)
出处
《电工技术》
2022年第23期176-178,共3页
Electric Engineering