With the development of the power system,the relationship between the electrical equipment becomes more complex,the requirement of controlling and protecting power system with the assistance of automatic survillance equipment is popular these days.Requirement mentioned above is reflected in controlling system,as well as in the status monitoring that consists of status shifting detection and malfunction warning ability.In this paper,a method based on video deshaking to detect switch cabinet when a moving camera stops after which extract subareas containing target components and analysis theirstatuses by particular image processing algorithms.To reduce the side effect of camera shaking and swinging caused by mechanical flaw of hardware,this paper proposes utilize the spatial transform of the feature key points in two adjcacent frames to smooth camera shaking and ensure detection process is triggered on a stable video,further detect the area of interesting covered by switch cabinet and perform corresponding algorithms to collect status data in pre-divided subareas.Experiments demonstrate that the proposed switch cabinet surveillance method has the advantages of camera shaking immunity,high efficiency,localization accuracy and robustness in varied scenes and environments.%随着电力系统的发展,电力设备之间的关系变得越来越复杂,通过自动化设备来控制和保护电力系统的方法已经广泛应用于电气行业中,对设备自动化的要求也越来越高.该类要求不仅体现在设备控制方面的自动化,监控设备也要具备自动识别状态变化和自动预警的功能.监控相机在拍摄时往往由于机械原因产生抖动和摇摆,提出了一种基于视频去抖动的电气开关柜状态自动识别方法,在视频中识别开关柜面板,并将其作为感兴趣区域,对开关柜上不同子部件进行划分和检测.为了减小因相机在拍摄时由于机械原因产生的抖动和摇摆,本文利用了检测前后两帧图像中特征点的运动变化,平滑相机抖动,减小软件误检的概率,并对开关柜上子部件实施针对性的检测算法,提高准确率.实验结果表明,提出的电气开关柜监控算法具有不受相机晃动干扰,实时性强、检测准确以及抗干扰能力强的优点.
展开▼
