基于双目视觉的混凝土电杆立杆参数测量方法

Measurement method of parameters of concrete electric pole based on binocular vision

为准确掌握混凝土电杆立杆参数,及时对抗倾覆能力不足的混凝土电杆进行加固以预防电杆倾倒事故,提出一种基于双目视觉的混凝土电杆埋深与倾斜率方法。首先通过修改主干特征提取网络、优化特征融合并引入改进CBAM注意力机制的Deeplab V3+语义分割模型,并基于采集的混凝土电杆图片数据集进行训练,确定混凝土电杆在全景图像中的区域;然后,采用基于局部区域效应的亚像素边缘检测算法提取分割区域内的混凝土电杆边缘点,并根据混凝土电杆的几何特性使用最小二乘法拟合边缘获取所需特征点,最后利用双目视觉的坐标变换计算混凝土电杆埋深与倾斜率。结果表明:该方法的埋深测量误差值小于10 cm,误差率小于5%,倾斜率测量误差值小于0.3°,程序运行耗时小于4 s,该方法在混凝土电杆立杆参数测量问题上具有可行性,测量精度与测量速度满足实际应用需要。

In order to accurately determine the parameters of the concrete electric pole and strengthen the concrete electric pole with insufficient overturning ability in time to prevent the pole dumping accident,a method of buried depth and inclination rate of concrete electric pole based on binocular vision is proposed.Firstly,the backbone feature extraction network was modified,the feature fusion was optimized,and the Deeplab V3+semantic segmentation model with improved CBAM attention mechanism was introduced,as well as the region of the concrete pole in the panoramic image was determined by training based on the collected concrete pole image dataset.Then,the sub-pixel edge detection algorithm based on local area effect was used to extract the edge points of the concrete power pole in the segmented region,and the least square method was adopted to fit the edge according to the geometric characteristics of the concrete power pole to obtain the required feature points.Finally,the coordinate transformation of binocular vision was used to calculate the buried depth and inclination rate of the concrete power pole.The results show that the error value of buried depth measurement is less than 10 cm,the error rate is less than 5%,the error value of inclination rate measurement is less than 0.3,and the running time of the program is less than 4 s.The method is feasible in the measurement of the concrete pole parameters,and the measurement accuracy and measurement speed could meet the needs of practical applications.