基于双目视觉的电力塔倾斜检测方法  

Binocular vision-based tilt detection method for power towers

作　　者：陈广华[1] 葛梦莹 黄白瑶 梁国贤 李潇凯 CHEN Guanghua;GE Mengying;HUANG Baiyao;LIANG Guoxian;LI Xiaokai(School of Mechanical and Electronic Control Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学机械与电子控制工程学院,北京100044

出　　处：《北京交通大学学报》2022年第2期128-137,共10页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

基　　金：国家自然科学基金(51376017)。

摘　　要：电力塔倾斜检测对电力系统安全运行具有至关重要的作用.针对传统电力塔倾斜检测方法效率低的问题,设计了一种基于双目视觉的塔倾斜检测系统,该系统具有远程智能、准确且便捷的特点.通过张氏标定法进行双目摄像机的标定与校正,基于DeepLab V3+对电力塔图像区域进行分割并提出融合低阶特征的DeepLab V3+网络结构.由极线立体匹配算法得到电力塔侧棱视差,通过该侧棱特征点的三维坐标拟合侧棱空间直线方程,建立倾斜度计算模型估计电力塔倾斜度.最后和经纬仪测量结果进行对比,平均测量误差为0.020°,在技术标准的允许误差之内,实验说明该系统可以准确便捷地计算出电力塔倾斜度.The tilt detection for power tower plays a critical role in ensuring the safe operation of power system. To solve the problem of low efficiency of traditional tilt detection method for power tower, a binocular vision-based tower tilt detection system is designed. The system is characterized by remote intelligence, accuracy and convenience. Zhang’s Calibration method is used to calibrate and correct the binocular camera. Deeplab V3+ is used to segment the image region of power tower, and a Deeplab V3+ network structure is proposed to fuse low-order features. The parallax of the side prism of the power tower is obtained by the polar line stereo matching algorithm, and the l spatial linear equation of the side prism is fitted by the 3D coordinates of the side sprism feature points, and the tilt calculation model is established to estimate the tilt of the power tower. Finally, after comparing with the longitudinal measurement results, the average measurement error is 0.020°, which is within the allowable error of the technical standard. The experiment shows that the system can accurately and conveniently calculate the tilt of power tower.

关 键 词：电力塔倾斜度 双目立体视觉系统 特征提取 立体匹配 倾斜度计算 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置] TP39[自动化与计算机技术—控制科学与工程]