基于视觉定位技术的水力发电厂入侵人员UWB定位方法

UWB Location Method of Intruders in Hydropower Plant Based on Vision Location Technology

为了实现复杂环境下入侵人员精准定位,研究开发了基于视觉定位技术的水力发电厂入侵人员UWB定位方法。该方法利用双目视觉技术,将采集到的入侵人员图像作为深度学习网络的输入,并在深度学习网络中引入用于多特征提取的金字塔网络层、用于空间上下文增强的空洞卷积层,实现水力发电厂入侵人员检测;利用优化UWB定位技术计算标签位置,完成水力发电厂入侵人员定位。该方法可清晰采集入侵人员图像,有效检测入侵人员,并完成不同干扰环境下的入侵人员精准定位,实时追踪入侵人员行动轨迹,并对入侵行为实施相应预警。

In order to achieve precise localization of intruders in complex environments,a UWB localization method for intruders in hydroelectric power plants based on visual localization technology is studied.Using binocular vision technology to collect images of intruders from hydroelectric power plants and taking the collected images of intruders as input to the deep learning network,a pyramid network layer for multi feature extraction and a hollow convolutional layer for spatial context enhancement are introduced to achieve intrusion detection in hydroelectric power plant.The intrusion detection results are treated as ultra wideband positioning tags,and the tag position is calculated using optimized UWB positioning technology to complete the localization of intrusion personnel in hydroelectric power plants.The method described in this article can capture intruders images clearly,detect intruders effectively and locate intruders accurately in different interference environments,and the same time,track the trajectory of intruders in real-time and implement corresponding warnings for intrusion behavior.