Research On Power Line Status And Fault Recognition Algorithm Based On Uav Aerial Photography

As an important part of power lines,power lines bear the important responsibility of power transmission.Due to its long-term exposure to natural environment,there will be faults such as broken stock and foreign matter hanging on the power line.If it is not found in time,it will have adverse effects on power lines.At present,the main power line inspection methods have manual inspection and helicopter inspection,these two ways of work intensity,easy to be affected by the weather,and later human screening failure image efficiency is low.Therefore,aiming at the aerial line image of UAV,this paper focuses on how to identify the broken wires and suspended foreign bodies in power lines.In view of the complex background and diverse noise characteristics of power line images,image preprocessing is studied.Using image grayscale,histogram equalization,and wavelet transform denoising technology,the quality of aerial images can be improved.In order to ensure the accuracy of power line extraction,edge detection is carried out first.Several commonly used edge detection algorithms are studied.The traditional Canny edge detection algorithm is improved in two aspects of computing gradient amplitude and threshold,and the accuracy and accuracy of the improved algorithm are improved.According to the straight line characteristics of power lines,a straight line detection algorithm is studied.The line detection algorithm based on one dimensional transformation of Hough is used to extract the power line.This algorithm can effectively solve the problem that the false straight line is easily produced in the traditional line detection algorithm.Aiming at the broken stock and foreign object suspension on power line,a fault recognition algorithm model is designed.The model corresponds to two fault recognition algorithms for two kinds of fault types,which can realize accurate identification of broken stock and foreign object in turn.To identify the broken stock,a cross point recognition algorithm based on local contour features is proposed.The algorithm first decomposes the overall template contour of the shape objects at the intersection point of the broken stock,and statistics the local contour features of all possible 2AS or 3AS shapes of the shape objects,then uses the 2AS descriptor of the semantic model to redefine the feature description of the template,and sets up the sample library through the K mean clustering,and finally the Chi-square method will be used.The matching image is matched with the template in the sample library to realize the recognition of broken stock.The width pixels of the power line are analyzed,and a foreign object recognition algorithm based on the width of the power line is proposed.By setting the threshold of the width of the power line,the target of the foreign object suspension is identified.The experimental results confirm that the algorithm model designed in this paper can identify the broken stock and foreign matter suspension accurately and efficiently.