Lightning disasters pose a serious threat to railway operations due to their strong spatiotemporal randomness and high destructiveness. Traditional lightning location methods rely on manual inspections, which suffer from difficulties in detection and delayed responses. To address the aforementioned challenges, a computer vision-based positioning method was proposed that integrated adaptive background subtraction with enhanced edge detection, specifically designed for complex railway scenes and overexposure issues in lightning imaging. This approach employed a dynamic Gaussian mixture model for background modeling, combined an improved Canny operator to enhance lightning channel feature extraction, and incorporated morphological filtering to eliminate noise interference, thereby establishing an efficient framework for lightning target detection. To verify the effectiveness of the method, 37 sets of natural lightning cases with different scenarios were selected for experimentation. The results show that the recall and accuracy of lightning channel recognition are 0.86 and 0.89, respectively, and the accuracy of lightning landing point recognition reaches 97.3%. The proposed method significantly improves the robustness and real-time performance of lightning localization in complex environments, effectively alleviates the problem of feature loss caused by overexposure, and has high accuracy in lightning landing point localization. It can provide reliable technical support for intelligent monitoring and rapid emergency response of railway lightning disasters.
XIAO Tong, XIE Zhi-duo, LI Hui, JIN Shao-yun, MA Ying.
Lightning Strike Location Method in Complex Scenarios Based on Computer Vision[J].
Science Technology and Engineering, 2026, 26(28): 12011-12021 DOI:
为了更好地实现复杂背景下闪电通道的识别与闪电落点的定位,并尽量贴合真实应用场景,本实验选用中国气象局雷电野外科学试验基地(Field Experiment Base on Lightning Sciences, China Meteorological Administration,CMA-FEBLS)提供的城市观测数据作为数据集的重要补充。重点选取背景复杂度高的图片,如包含多栋建筑物、光斑亮点、地物轮廓模糊等特点。广州高建筑物雷电观测站(Tall-object Lightning Observatory in Guangzhou,TOLOG)位于广州市中心区域,作为CMA-FEBLS的关键组成部分始建于2009年,迄今已积累数百次高建筑物雷电资料。该观测站包含6个观测点,主观测点东南方向视野范围内有8座高度超过300 m的建筑物,通过站内设置的闪电通道成像仪、高速摄像机和全视野通道成像仪等多种光学观测设备,能够全面地对广州珠江新城地区高建筑物上的闪电过程进行观测,清晰捕捉自然闪电击中建筑物的图像[20]。该观测站完备的观测设备,能够为本次实验提供清晰、翔实的数据,其复杂的背景环境较为贴近铁路站场,便于调试算法、验证效果。
XiaoTong, NiFenbin, JinShaoyun, et al. Research and application of railway lightning detection system based on hybrid cloud[J]. Railway Signalling & Communication, 2026, 62(2): 6-14.
FuMaojin, XiaoTong. Research and prospect of lightning protection technology for railway communication signals[J]. Railway Signalling & Communication, 2019, 55(S1): 167-170.
[7]
侯文玉. 高速铁路接触网雷害特征与防护[J]. 中国铁路, 2016(11): 1-6.
[8]
HouWenyu. Characteristics and protection of lightning damage in high-speed railway catenary[J]. China Railway, 2016(11): 1-6.
YangJun, LiuYan, LiJie, et al. Research on lightning type recognition based on multi-scale residual networks[J]. Science Techno-logy and Engineering, 2025, 25(27): 11483-11489.
[11]
WarnerT A. Observations of simultaneous upward lightning leaders from multiple tall structures[J]. Atmospheric Research, 2012, 117: 45-54.
[12]
GaoY, ChenM, QinZ, et al. The spatial evolution of upward po-sitive stepped leaders initiated from a 356-m-tall tower in southern China[J]. Journal of Geophysical Research: Atmospheres, 2020, 125(2): 1-18.
[13]
SuZ, LüW, ChenL, et al. Shielding effect of surrounding buildings on the lightning-generated vertical electric field at the top of a tall building[J]. IEEE Transactions on Electromagnetic Compatibility, 2019, 61(1): 174-182.
[14]
ChenL, LüW, ZhangY, et al. Correlated luminosity and magnetic field peaks produced by canton tower-strokes[J]. Atmospheric Research, 2019, 218: 59-69.
[15]
FanX, ZhangY, KrehbielP R, et al. Application of ensemble empirical mode decomposition in low-frequency lightning electric field signal analysis and lightning location[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(1): 86-100.
[16]
FanX, ZhangY, KrehbielP, et al. Channel development and electric parameter characteristics of regular pulse bursts in lightning[J]. Geophysical Research Letters, 2024, 51(7): e2023GL106582.
YangXinyi, LüWeitao, YangJun, et al. Applications of three threshold methods to the lightning channel image recognition[J]. Journal of Applied Meteorological Science, 2014, 25(4): 427-435.
GaoYan, LüWeitao, ChenLüwen, et al. Reconstruction method of 3D lightning channel from two-station video observation[J]. Plateau Meteorology, 2015, 34(3): 842-849.
LiFeng, LüWeitao, LiQingyong, et al. Lightning channel image recognition based on line support region[J]. Journal of Applied Meteorological Science, 2016, 27(6): 725-733.
KangXuedong. A preliminary exploration of intelligent construction and operation management of railways in China[J]. Journal of Railway Engineering Society, 2019, 36(4): 84-89.
LiXingxin, ZhuLiqiang, YuZujun. Adaptive foreground object detection in railway scene[J]. Journal of Transportation Systems Engineering and Information Technology, 2020, 20(2): 83-90.
WangFei, LiuGuiwei, ChenZelian, et al. Research on railway foreign object intrusion detection technology based on U-Net siamese network[J]. Journal of Railway Engineering Society, 2025, 42(4): 88-92.
China State Railway Group Co., Ltd. Technicalspecificationforintegratedvideomonitoringsystemofrailway:Q/CR575—2022[S]. Beijing: China Railway Publishing House, 2022.
LüWeitao, ChenLüwen, MaYing, et al. Advances of observation and study on tall-object lightning in Guangzhou over the last decade[J]. Journal of Applied Meteorological Science, 2020, 31(2): 129-145.
ZhaoYalin, TaoQingchuan, LuQianwen, et al. The Canny threshold selection algorithm based on the second derivative of image gradient[J]. Science Technology and Engineering, 2018, 18(14): 179-185.
[39]
NarayanaM, HansonA, Learned-MillerE G. Background subtraction: separating the modeling and the inference[J]. Machine Vision and Applications, 2014, 25(5): 1163-1174.
ShuZhaohan, LiXiaolong, GuoXia, et al. Improved ViBe algorithm for ghosting suppression and its application[J]. Science Technology and Engineering, 2025, 25(14): 5948-5956.
JiangXingyu, SongZhenqiang, XueGang, et al. Individual mango tree Canopy extraction by coupling multi-scale convolution and boundary detection techniques[J]. Science Technology and Engineering, 2025, 25(33): 14371-14379.
DuWenhan, LiDongxing, WangQiannan, et al. Moving target detection based on improved frame difference and edge extraction algorithm[J]. Science Technology and Engineering, 2022, 22(5): 1944-1949.
FengPengfan, HaoRujiang, ShangTenglong, et al. Gearbox fault diagnosis based on MDF and efficientnet[J]. Science Technology and Engineering, 2026, 26(3): 1038-1045.
[52]
YoshinagaS, ShimadaA, NagaharaH, et al. Object detection based on spatiotemporal background models[J]. Computer Vision and Image Understanding, 2014, 122: 84-91.
LuoWei, WuChaohua, XiaoJun, et al. Three-threshold image segmentation of welding defects based on improved particle swarm algorithm[J]. Science Technology and Engineering, 2025, 25(22): 9463-9470.
CaiJing, WangKai, LiYue, et al. Infrared image airport runway water identification method based on the fusion algorithm of Canny operator + fuzzy C-means clustering[J]. Science Technology and Engineering, 2024, 24(28): 12382-12390.
[57]
ZhangK Q, ChenS C, WhitmanD, et al. A progressive morphological filter for removing nonground measurements from airborne LIDAR data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2003, 41(4): 872-882.
[58]
FelzenszwalbP F, GirshickR B, McAllesterD, et al. Object detection with discriminatively trained part-based models[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(9): 1627-1645.