Invasion Detection Research Of Intelligent Surveillance Electric System

Nowadays,an intelligent surveillance system for invasion detection of transmission lines using video images from aerial patrol helicopter has shown advantages of high safety and reliability,geographical invariance,low cost and high efficiency.High voltage electrical transmission system usually consists of two parts: pylons and transmission lines,on both of which malfunction tend to occurs.Invasions on these two parts bring great threat to the safe operation of the whole electric system.We solve these two problem separately with different methods based on computer vision technology.For pylons,we propose a detection method for invasions on the pylons based on a coarse to fine searching strategy.In order to get the location of pylons accurately and rapidly,we firstly use the classifier to search the pylons in the images and then employ the tracking method to consecutively locate the pylons.Considering that the structures of pylons are very thin and difficult to be distinguished from complex backgrounds,we propose to establish a classifier based on sparse features of pylons to achieve promising performance of pylon detecting and tracking,which is illustrated to have good invariance to view changes and cluttered background.Given the location of pylons,we can further find out a candidate region of malfunction based on color filtering.As compared with the traditional methods,the experimental results demonstrate that the proposed pylon net detection method has the advantages of high efficiency,localization accuracy and robustness in varied scenes.For transmission lines,we propose a method based on LSD(Line Segment Detector)to extract transmission lines and form ROI(Regions of Interest).Considering that invasion detection for power transmission line isan important issue for public safety and power supply this paper proposes to realize general invasion detection using HVS(Human Vision System)saliency analysis,which could efficiently distinguish invasion object with its different appearance from transmission lines in color,shape,and spatial distribution.Specifically,the method calculates the saliency mask around the pre-acquired transmission lines to achieve real-time invasion detection.The experimental results demonstrate that the proposed invasion detection method has advantages of high location accuracy and strong robustness in complex scenes.