基于无人机红外热成像的架空输电线视觉跟踪巡检研究

发布时间：2018-09-10 14:55

【摘要】：随着技术的进步,近年来无人机在架空输电线路巡检中占据了非常重要的地位。国内外围绕无人机拍摄影像的自动检测处理展开了大量的研究和应用,但是巡检时输电线目标极容易脱离镜头画面,造成漏检问题。为了使无人机巡检更具有实用性,影像的智能采集是研究的关键,必须解决无人机的长距离自主巡航以及巡航过程中机载相机准确跟随线路拍摄等问题。本文针对这一问题提出了一种基于空间视觉检测的新型巡检方案,依托卫星导航的长航时多旋翼无人机搭载红外相机和可见光相机同时拍摄线路,红外相机则负责跟踪输电线路,通过对线路特征较清晰的红外图像进行分析处理,输出控制信号及时调整云台的角度,保证输电线目标不会移出镜头画面。可见光相机同步拍摄线路、金具、绝缘子等细节照片,达到高质量巡检的目的。文中采用权值窗口优化的光流法进行红外图像处理,特别研究了杆塔转角时出现的短暂不等距斜视问题,提出了基于杆塔标志校正的方法,且对所涉及算法进行了分析与实验验证。提出了一种基于k-b参数空间的模板集合匹配对直线检测和拟合的图像处理算法,提高了对线路和杆塔的提取清晰度。完成了云台角度调整的理论验证。本文针对上述问题搭建了硬件实验平台,制作八旋翼无人机、可见光相机、红外相机以及云台、飞控等设备,进行了相关的理论验证和实验分析。通过仿真实验发现,上述方法可以有效的保证无人机搭载相机对输电线巡检图像的采集,在没有不可控因素出现的条件下,能够很好捕获输电线和杆塔,且计算有效的云台调整参数,对输电线路实现较好的跟踪,从而为无人机视觉追踪巡检提供一定理论参考。
[Abstract]:With the development of technology, UAV plays an important role in overhead transmission line inspection in recent years. A large number of researches and applications have been carried out around the automatic detection and processing of UAV images at home and abroad, but transmission line targets are very easy to detach from the lens picture during patrol inspection, which results in the problem of missing detection. In order to make the UAV patrol more practical, the intelligent image acquisition is the key to the research. It is necessary to solve the long distance autonomous cruise of UAV and the accurate track shooting of the airborne camera during the cruise. In order to solve this problem, a new inspection scheme based on space vision detection is proposed in this paper. The multi-rotor UAV based on satellite navigation is equipped with infrared camera and visible light camera to shoot the line simultaneously. The infrared camera is responsible for tracking the transmission lines. By analyzing and processing the infrared images with clear line features, the output control signal adjusts the angle of the cloud head in time, so as to ensure that the transmission line targets will not move out of the lens. Visible-light camera synchronously takes details such as lines, hardware, insulators and so on to achieve high-quality inspection. In this paper, the weight window optimized optical flow method is used for infrared image processing, especially the problem of transient non-isometric squint in the rotation angle of the tower is studied, and a method based on pole and tower marking correction is proposed. The algorithm is analyzed and verified by experiments. An image processing algorithm based on k-b parameter space is proposed to detect and fit straight lines by template set matching, which improves the clarity of line and tower extraction. The theoretical verification of the angle adjustment of the cloud head is completed. In order to solve the above problems, this paper builds a hardware experimental platform, and makes eight rotary-wing UAV, visible light camera, infrared camera, cloud head, flight control and other equipment, and carries out relevant theoretical verification and experimental analysis. Through simulation experiments, it is found that the above method can effectively ensure that the UAV can carry the camera to collect the image of the transmission line patrol inspection, and can capture the transmission line and tower very well without any uncontrollable factors. The effective parameters of cloud head adjustment are calculated, and the transmission line is tracked well, which provides a certain theoretical reference for the visual tracking inspection of UAV.
【学位授予单位】：兰州理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM755

【参考文献】