UAV Overhead Wire Image Recognition And Display System With Infrared Camera

With the rapid development of technologies such as UHV and high-voltage transmission,the scale of the State Grid is also growing.As one of the indispensable and important components in the national power system,overh ead conductors need to be inspected regularly to ensure the safe operation of the entire power grid s ystem.On the one hand,the detection of overhead conductors is the premise and basic work for completing tasks such as the integrity of overhead conductors,icing conditions,foreign matter attachment,tree and other vegetation intrusion,and drawing of electrical components on overhead conductors.On the other hand,the accurate and fast overhead wire extraction algorithm can be used to sense the range of overhead wires,assist UAVs to avoid obstacles,and play a role in the route formulation and navigation of UAVs and other equipment used for inspection.Key role.In this paper,the infrared images of overhead conductors are obtained by the DJ I Royal 2 ind ustrial UAV,and the deep learning and system design are combined to complete the rapid and real-time identification and tracking of overhead conductor targets in infrared images and infrared video streams.Real-time display,the main research content and results of this paper are as follows:The extraction method of overhead wires in infrared images is studied.This paper proposes a method for identifying overhead wire targets in infrared images using Deeplabv3+ semantic segmentation model,and improves th e algorithm according to the characteristics of wire targets,introduces feature pyramid and attention mechanism,and conducts anal ysis on porous space pyramid structure.Optimization to improve the identification accuracy of overhead conductors.The experimental results show that the improved algorithm proposed in this paper improves the pixel accuracy by 1.5%,the average pixel accuracy by 7.3%,and the average intersection ratio by 5.8% based on the original algorithm,which can effectively extract wire targets.The detection and tracking methods of overhead conductors in infrared video streams are studied.In this paper,detection-based tracking algorithm is used to complete the detection and tracking of overhead wires in infrared video streams.After completing the wire extraction from the infrared image using the Deeplabv3+ semantic segmentation algorithm,the video stream is decoded into a picture stream of continuous frames,and the trained Deeplabv3+ model is used to identify the overhead wires,and then use the optical flow method to predict the next frame.position,and use the Hungarian algorithm to complete the identification of overhead conductors and match and update the tracking results.Experiments show that using the improved Deeplabv3+ algor ithm as the detector,compared with the original Deeplabv3+ algorithm as the detector,the accuracy of the tracking algorithm is increased by 11%,and the success rate is increased by21%.Simulation design and implementation of real-time detection and dis play system.By means of system simulation,infrared image wire identification and real-time detection and tracking of wire targets in infrared video streams are realized,and displayed through a visual interface.The video rate collected in this paper is 29.97 frames per second,while the frame rate of the tracking algorithm can reach 30 frames per second,which can realize real-time detection and tracking of wire targets in infrared video streams.