Design of Transmission Line Identification and Localization System Based on Machine Vision

High-voltage transmission lines are often inspected by drones or power robots; the use of drones instead of manually lifting power robots onto power lines can save a lot of time and cost; in recent years, accidents of helicopters or drones colliding with power lines have occurred frequently, resulting in casualties and property losses. Whether it is the intelligent operation on power lines or the transmission line warning mechanism of low-altitude aircraft, a set of identification and positioning system for transmission lines is needed. Traditional vision algorithms have low accuracy in detecting slender transmission lines in complex scenes, and are greatly interfered by environmental factors; existing deep learning-based semantic segmentation algorithms for power lines are not efficient. To address these issues, this paper develops a transmission line identification and localisation software system. The system integrates the PowerLine Net algorithm model and binocular vision algorithm, and has the functions of wire detection and wire positioning. Finally, the functions of the system are verified through experiments, and it has certain application value.