基于改进YOLO与HSV变换的高压线缆识别定位方法

High-Voltage Cable Identification and Location Method Based on Object Detection and HSV Transformation

带电作业机器人在配电网维护领域可以有效解决传统维护手段工作强度大、安全风险高以及作业效率低等问题。针对带电作业机器人在作业过程中对高压线缆识别定位精度不高的问题,提出了一种基于改进YOLO与HSV变换相结合的识别定位方法。该方法首先对YOLO v4特征提取主干网络CSPDarknet53进行简化,并对空间金字塔池化网络(Spatial Pyramid Pooling,SPP)进行改进,提出CSP-SPP特征融合模块。其次结合HSV色彩追踪技术,提取ROI区域内线缆目标最小外接矩形。最后结合深度相机求得线缆目标上多个等分点三维坐标信息,并利用坐标信息完成对机械臂的引导。实验结果表明,提出的算法具有较高的识别速度和精度,可以有效地为带电作业机器人抓取线缆提供三维坐标信息,其中改进后的YOLO算法在测试集上的识别率达到97.38%。

In the field of power distribution network maintenance,live-working robots can effectively solve the shortcomings of traditional maintenance methods such as high work intensity,high security risk,and low operating efficiency.Aiming at the problem of low accuracy in intelligent recognition and location of high-voltage cables during the operation of live-working robots,a method of recognition and location based on the combination of improved YOLO and HSV transformation is proposed.The method first simplified the Yolo v4 feature extraction backbone network CSPDarknet53 and improved the Spatial Pyramid Pooling(SPP)network,then roposed CSP-SPP feature fusion module.The smallest bounding rectangle of the cable target in the ROI area are extracted combined with HSV color tracking technology.Finally,Multiple three-dimensional coordinate point informations on the cable target are obtained combined the depth camera,and the robotic arms are guided by the coordinate informations.The experimental results show that the algorithm proposed in this paper has higher recognition accuracy and speed,and can effectively provide three-dimensional coordinate point informations for the robot arms to grab the cables.The improved YOLO algorithm achieves a recognition rate of 97.38%on the test set.