Unmanned Aerial Vehicle Aerial Image Basis Transmission Lines Extraction And Tracking Method Research

Electrical energy is a very important energy form in human society,and transmission lines is the core for the entire power transmission and exchange.However,with the expansion of the transmission grid,the task of regular checking and inspections for the transmission lines became more and more onerous.In order to further reduce the workload,intelligent unmanned aerial vehicle(UAV)patrol inspection technology has become a hot research topic.Based on the UAV image,this article studies the transmission lines extraction and connection,and carries out automatic tracking simulation of the transmission lines,which lays a foundation for the future application of UAV for automatic route navigation and the establishment of the 3D model of the power corridor.The main research work and conclusion of this article as follows:(1)Designed the UAV patrol image acquisition scheme.In the test area with length about100 m,327 patrol images and 4 episodes of patrol videos were collected through two times of experiments.Directed at the problem of low contrast in the image due to the influence of light during UAV shooting,the operand is reduced by graying,and the image enhancement processing is executed by histogram equalization;aimed at problem of the complex background and the large mixed noise of the UAV image,median filtering and bilateral filtering were respectively used,the gray histogram and peak signal-to-noise ratio(PSNR)results shown that the bilateral filtering was suitable to filter the mixed noises with good edge preservation ability.(2)Used the first-order differential algorithm,Canny algorithm and Line Segment Detector(LSD)algorithm for the boundary extraction of the pretreatment image.Combined with the features of the transmission lines image,the LSD algorithm was optimized to make the extracted transmission lines edges more complete and reduced the miss detection of lines edge.Adopted the connected components analysis method to remove the background interference short line segments from the boundary extraction result.Used the Hough transform method to detect the lines to obtain a linear equation,linearly grouped the transmission lines and the tower by the slope to determine the linear cluster of the towers,and accurately connected the transmission lines at last.(3)Used particle filter and extended Kalman filter(EKF)these two tracking methods to build the motion model of the transmission lines in adjacent frames.The position of the transmission line in the current video frame can be used to predict the ROI area in the next frame of transmission lines,narrowing the search scope for the lines extraction,in this case,the identification efficiency of transmission lines was improved.In the MATLAB simulation results,the inspection accuracy of particle filter algorithm was 94.21%,while that of EKF algorithm was 93.68%,with little difference between them.The application of the EKF algorithm would not only satisfies the tracking requirements,but also improves the real-time detection.