基于无人机图像的电力杆塔倾斜检测  

The method of Power Transmission Tower Inclination Detection Based on UAV Image

作　　者：王榆夫 韩军[1] 赵庆喜 王亚先[1] 

机构地区：[1]上海大学通信与信息工程学院,上海200444 [2]河南立新监理咨询有限公司,河南郑州450052

出　　处：《计算机仿真》2017年第7期426-431,共6页Computer Simulation

基　　金：国家自然科学基金面上项目(61471230);面向无约束视频的时空显著性模型及其应用研究(2015/01-2018/12)

摘　　要：针对航拍图像中电力杆塔倾斜的自动识别,提出了一种基于杆塔结构知识模型和航拍方案相结合的杆塔姿态倾斜检测算法。首先,对图像中的线段特征进行检测和提取。其次,借助基于杆塔结构知识模型的交叉点分析法对线段进行分类同时对杆塔进行识别,提取能够表征杆塔的主要轮廓,并根据主要轮廓进行中轴的提取。最后,根据航拍成像的特点,提出一种简单的电力杆塔巡检航拍方案。并依据所提出的方案选取地面法线的参考方向,根据地面法线参考方向与杆塔中轴的夹角对杆塔的倾斜进行判决。实验证明,改进方法能够有效识别杆塔,并对杆塔的倾斜进行检测。In order to automatically identify whether a tower is skewed in the aerial image, a novel algorithm whiteh combined the knowledge of tower structure and the characteristics of aerial photography is proposed in this pa- per. Firstly, the features of line segments in the image were detected and extracted. Secondly, with the help of knowledge model of tower, line segments were classified by means of cross point analysis and the electric tower was i- dentified at the same time. And the main contours were extracted to calculate the main axis. Finally, according to the characteristics of aerial images, a simple shooting scheme of power tower inspection was proposed. The reference di- rection of the ground normal was selected based on the proposed scheme and the angle between the selected reference direction and the extracted axis of the transmission tower was used to determinate whether the tower is tilt. Experi- ments show that this method can effectively identify the transmission tower in aerial images, and to detect the inclina- tion of the tower.

关 键 词：航拍图像 交叉点分析 电力杆塔识别 倾斜检测 

分 类 号：TN911.73[电子电信—通信与信息系统]