斯建东,汤义勤,赵文浩
SI Jiandong,TANG Yiqin,ZHAO Wenhao

• 1:国网浙江省电力有限公司台州供电公司

摘要(Abstract)：

针对目前无人机搭载传感器的树障检测方法无法实现自动检测的问题，提出一种基于改进的FPN（特征金字塔网络）与SVM（支持向量机）的树障检测算法。在传统的FPN基础上，进行自下而上的反向侧边连接并融合，采用ResNet 50（深度残差网络）和改进的FPN作为特征提取网络得到特征向量，并将其输入基于遗传算法的SVM中进行二分类，进而判断所检测图像中是否存在树障隐患。实验结果表明，本算法用于树障检测的准确率达到93.4%，处理图像的平均速度达到每秒11张，漏检率和误检率较低，具有较强的泛化能力。
The current tree barrier detection methods with UAV-mounted sensors cannot realize automatic detection.Therefore, this paper proposes a tree barrier detection algorithm based on improved FPN(feature pyramid network) and SVM(support vector machine). On this basis of traditional FPN, bottom-up reverse side connection and fusion is performed. ResNet 50, the deep residual networks, and the improved FPN are used as feature extraction networks to obtain the feature vectors, which are inputted into SVM based on genetic algorithm to perform binary classification and then determine whether there is a tree barrier hidden in the detected image. The experimental results show that the accuracy of this algorithm for tree barrier detection reaches 93.4%, and the average speed of processing images reaches 11 images per second, characterized by low missing detection rate and false detection rate as well as strong generalization ability.

关键词(KeyWords)： 无人机;树障检测;特征金字塔;深度残差网络;支持向量机
UAV;tree barrier detection;FPN;deep residual networks;SVM

Abstract:

Keywords:

基金项目(Foundation): 国网浙江省电力有限公司科技项目（5211TZ2000M7）

作者(Author): 斯建东,汤义勤,赵文浩
SI Jiandong,TANG Yiqin,ZHAO Wenhao

DOI: 10.19585/j.zjdl.202309015

参考文献(References)：

• [1]刘正坤，陈伦清，王昊.无人机辅助电网巡检作业的应用现状与思考[J].南方能源建设，2017,4(2):115-119.LIU Zhengkun,CHEN Lunqing,WANG Hao. Application status and reflections of electrical network inspection aided by unmanned aerial vehicle[J]. Southern Energy Construction,2017,4(2):115-119.
• [2]曾忱，邹俊，林月峰.基于无人机可见光点云的输电线路树障隐患智能分析研究[J].电力设备管理，2020(3):38-41.ZENG Chen,ZOU Jun,LIN Yuefeng.Intelligent analysis and research on hidden danger of transmission line tree obstacle based on UAV visible light point cloud[J].Electric Power Equipment Management,2020(3):38-41.
• [3]成广生，罗培焱，刘志武，等.基于多光谱和LiDAR数据融合技术的树种分布识别及树障分级预警系统研究[J].东北电力技术，2021,42(3):29-32.CHENG Guangsheng,LUO Peiyan,LIU Zhiwu,et al.Research on tree species distribution identification and tree obstacle classification early warning system based on multi-spectral and LiDAR data fusion technology[J].Northeast Electric Power Technology,2021,42(3):29-32.
• [4]王红星，陈玉权，张欣，等.基于离线高斯模型的输电线路无人机巡检缺陷智能识别方法研究[J].电测与仪表，2022,59(3):92-99.WANG Hongxing,CHEN Yuquan,ZHANG Xin,et al.Research on intelligent recognition method of transmission line UAV inspection defects based on offline Gaussian model[J]. Electrical Measurement&Instrumentation,2022,59(3):92-99.
• [5]魏业文，李梅，解园琳，等.基于改进Faster-RCNN的输电线路巡检图像检测[J].电力工程技术，2022,41(2):171-178.WEI Yewen,LI Mei,XIE Yuanlin,et al. Transmission line inspection image detection based on improved FasterRCNN[J].Electric Power Engineering Technology,2022,41(2):171-178.
• [6]沙伟燕，何宁辉，丁培，等.基于无人机图像处理的输电线路提取技术研究[J].电测与仪表，2022,59(5):158-165.SHA Weiyan,HE Ninghui,DING Pei,et al.Research on transmission line extraction technology based on UAV image processing[J].Electrical Measurement&Instrumentation,2022,59(5):158-165.
• [7] GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition. June 23-28,2014,Columbus,OH,USA.IEEE,2014:580-587.
• [8] HE K M,ZHANG X Y,REN S Q,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(9):1904-1916.
• [9] REN S Q,HE K M,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
• [10] HE K M,GKIOXARI G,DOLLáR P,et al. Mask RCNN[C]//2017 IEEE International Conference on Computer Vision(ICCV).October 22-29,2017,Venice,Italy.IEEE,2017:2980-2988.
• [11] LIN T Y,DOLLáR P,GIRSHICK R,et al.Feature pyramid networks for object detection[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). July 21-26,2017,Honolulu,HI,USA. IEEE,2017:936-944.
• [12] CORTES C,VAPNIK V. Support-vector networks[J].Machine Language,1995,20(3):273-297.
• [13] LECUN Y,BOTTOU L,BENGIO Y,et al. Gradientbased learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
• [14] SIMONYAN K,ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL].(2015-04-10)[2022-11-02]. https：//arxiv. org/abs/1409.1556.
• [15] SZEGEDY C,LIU W,JIA Y Q,et al.Going deeper with convolutions[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).June 7-12,2015,Boston,MA,USA.IEEE,2015:1-9.
• [16]王海燕，张渺，刘虎林，等.基于改进的ResNet网络的中餐图像识别方法[J].陕西科技大学学报，2022,40(1):154-160.WANG Haiyan,ZHANG Miao,LIU Hulin,et al.Chinese food image recognition method based on improved ResNet[J].Journal of Shaanxi University of Science&Technology,2022,40(1):154-160.
• [17]李莉，乔璐，张浩洋.结合FPN改进R-FCN的肺结节检测算法[J].计算机应用与软件，2022,39(4):179-184.LI Li,QIAO Lu,ZHANG Haoyang.Improved r-fcn lung nodule detection algrithm based on fpn[J].Computer Applications and Software,2022,39(4):179-184.
• [18] PLATT J. Sequential minimal optimization:a fast algorithm for training support vector machines[EB/OL].(1998-04-21)[2022-11-03]. https：//www. semanticscholar. org/paper/Sequential-Minimal-Optimization-%3A-A-Fast-Algorithm-Platt/53fcc056f79e04daf11eb798a7238e93699665aa.
• [19]洪期望，李捍东.基于支持向量机的手势识别研究[J].微处理机，2022,43(2):47-50.HONG Qiwang,LI Handong.Research on gesture recognition based on support vector machine[J].Microprocessors,2022,43(2):47-50.
• [20]李丹.基于遗传算法的自适应学习路径探究[J].电脑编程技巧与维护，2022(3):50-51.LI Dan. Research on adaptive learning path based on genetic algorithm[J].Computer Programming Skills&Maintenance,2022(3):50-51.
• [21] SELVARAJU R R,COGSWELL M,DAS A,et al.GradCAM:visual explanations from deep networks via gradient-based localization[C]//2017 IEEE International Conference on Computer Vision(ICCV).October 22-29,2017,Venice,Italy.IEEE,2017:618-626.