基于全天空视频数据的雷击方位角自动识别方法
Lightning Azimuth Image Recognition Based on the All-Sky Deep Video Data
DOI: 10.12677/orf.2024.146579, PDF,   
作者: 孙海洋, 许丽人：地理信息工程国家重点实验室，陕西 西安；王家磊：南京信息工程大学气象灾害预警预报与评估协同创新中心/气象灾害重点实验室，江苏 南京；彭晓光, 徐 洋：北京蓝湖空间科技发展中心，北京
关键词: 全天空；自动识别；方位角；边缘检测法；Lightning Detection； Automatic Detection； Deep Neural Network； Image Recognition
摘要: 光学雷电探测是一种常用的手段，但通常是通过人为值守的方式进行，因此，如何实现闪电光学图像及其方位角的自动识别是非常重要的，因为这样可以大大降低人为值守的工作量。本文收集了大量的闪电图像数据，建立了包含超过三万个带标签样本的数据集。同时，根据闪电具有显著时变特征的性质提出了一个时序合成的预处理方法，结果表明时序合成的预处理方法大幅提高了模型分类准确率。训练出的模型实现了“是闪电”和“非闪电”的分类，在测试集上能达到98.6%的正确率。另外，本文发现，对有些闪电图像，基于边缘检测算法可以提取其闪电通道边缘，获得比较精确的角度信息。而对边缘检测算法失效的图像，可以采取垂直亮度累加法获得其大致角度范围。
Abstract: Lightning optical detection is an important method, however, it is often used to observe the lightning discharge based on the manual duty. Therefore, how to realize the automation lightning optical detection is very important, which can remarkably reduce our workload. In this paper, we have built a lightning image dataset, which contains more than 30,000 labeled samples, including two categories of “lightning” and “non-lightning”. At the same time, the data is preprocessed according to the characteristics of the lightning image recognition task, so that the model can obtain the time-varying characteristics of lightning, and the classification accuracy of the model is greatly improved. Through experimental verification, using DenseNet161 as the backbone network can achieve 86.5% recall rate and 0.2% false detection rate on the test set. Also, it is found that the edge detection method is well used for the image recognition of the lightning channel and its strike point, and for the lightning images that the edge detection method cannot identify, the vertical brightness accumulation method is used for the approximate azimuth range recognition.
文章引用：孙海洋, 许丽人, 王家磊, 彭晓光, 徐洋. 基于全天空视频数据的雷击方位角自动识别方法[J]. 运筹与模糊学, 2024, 14(6): 798-810. https://doi.org/10.12677/orf.2024.146579

参考文献

[1] Schonland, B.F.J. and Collens, H. (1934) Progressive Lightning. Transactions of the South African Institute of Electrical Engineers, 25, 124-135.
[2] Yokoyama, S., Miyake, K., Suzuki, T. and Kanao, S. (1990) Winter Lightning on Japan Sea Coast-Development of Measuring System on Progressing Feature of Lightning Discharge. IEEE Transactions on Power Delivery, 5, 1418-1425. [Google Scholar] [CrossRef] 
[3] Jiang, R., Wu, Z., Qie, X., Wang, D. and Liu, M. (2014) High‐Speed Video Evidence of a Dart Leader with Bidirectional Development. Geophysical Research Letters, 41, 5246-5250. [Google Scholar] [CrossRef] 
[4] Pu, Y., Qie, X., Jiang, R., Sun, Z., Liu, M. and Zhang, H. (2019) Broadband Characteristics of Chaotic Pulse Trains Associated with Sequential Dart Leaders in a Rocket‐Triggered Lightning Flash. Journal of Geophysical Research: Atmospheres, 124, 4074-4085. [Google Scholar] [CrossRef] 
[5] Jiang, R., Srivastava, A., Qie, X., Yuan, S., Zhang, H., Sun, Z., et al. (2021) Fine Structure of the Breakthrough Phase of the Attachment Process in a Natural Lightning Flash. Geophysical Research Letters, 48, GL091608. [Google Scholar] [CrossRef] 
[6] Sun, Z., Qie, X., Jiang, R., Liu, M., Wu, X., Wang, Z., et al. (2014) Characteristics of a Rocket‐Triggered Lightning Flash with Large Stroke Number and the Associated Leader Propagation. Journal of Geophysical Research: Atmospheres, 119, 388-399. [Google Scholar] [CrossRef] 
[7] Wang, C.-X., Qie, X.-S., Jiang, R.-B. and Yang, J. (2012) Propagating Properties of a Upward Positive Leader in a Negative Triggered Lightning. Acta Physica Sinica, 61, Article 039203. [Google Scholar] [CrossRef] 
[8] Kong, X., Qie, X. and Zhao, Y. (2008) Characteristics of Downward Leader in a Positive Cloud‐to‐Ground Lightning Flash Observed by High‐Speed Video Camera and Electric Field Changes. Geophysical Research Letters, 35, L05806. [Google Scholar] [CrossRef] 
[9] 李宗祥, 蒋如斌, 吕冠霖, 刘明远, 孙竹玲, 张鸿波, 刘昆, 李小强, 张雄. 人工引发闪电上行负先导的发展传输特征[J]. 物理学报. 2021, 70(19): 338-351.
[10] Chen, L., Lu, W., Zhang, Y. and Wang, D. (2015) Optical Progression Characteristics of an Interesting Natural Downward Bipolar Lightning Flash. Journal of Geophysical Research: Atmospheres, 120, 708-715. [Google Scholar] [CrossRef] 
[11] Lu, W., Chen, L., Zhang, Y., Ma, Y., Gao, Y., Yin, Q., et al. (2012) Characteristics of Unconnected Upward Leaders Initiated from Tall Structures Observed in Guangzhou. Journal of Geophysical Research: Atmospheres, 117, D19211. [Google Scholar] [CrossRef] 
[12] Lu, W., Wang, D., Takagi, N., Rakov, V., Uman, M. and Miki, M. (2008) Characteristics of the Optical Pulses Associated with a Downward Branched Stepped Leader. Journal of Geophysical Research: Atmospheres, 113, D21206. [Google Scholar] [CrossRef] 
[13] Lu, W., Wang, D., Zhang, Y. and Takagi, N. (2009) Two Associated Upward Lightning Flashes That Produced Opposite Polarity Electric Field Changes. Geophysical Research Letters, 36, L05801. [Google Scholar] [CrossRef] 
[14] Lu, W., Chen, L., Ma, Y., Rakov, V.A., Gao, Y., Zhang, Y., et al. (2013) Lightning Attachment Process Involving Connection of the Downward Negative Leader to the Lateral Surface of the Upward Connecting Leader. Geophysical Research Letters, 40, 5531-5535. [Google Scholar] [CrossRef] 
[15] Gao, Y., Lu, W., Ma, Y., Chen, L., Zhang, Y., Yan, X., et al. (2014) Three-Dimensional Propagation Characteristics of the Upward Connecting Leaders in Six Negative Tall-Object Flashes in Guangzhou. Atmospheric Research, 149, 193-203. [Google Scholar] [CrossRef] 
[16] Lu, W., Zhang, Y., Zhou, X., Qie, X., Zheng, D., Meng, Q., et al. (2009) Simultaneous Optical and Electrical Observations on the Initial Processes of Altitude-Triggered Negative Lightning. Atmospheric Research, 91, 353-359. [Google Scholar] [CrossRef] 
[17] Zhou, E., Lu, W., Zhang, Y., Zhu, B., Zheng, D. and Zhang, Y. (2013) Correlation Analysis between the Channel Current and Luminosity of Initial Continuous and Continuing Current Processes in an Artificially Triggered Lightning Flash. Atmospheric Research, 129, 79-89. [Google Scholar] [CrossRef] 
[18] Wang, J., Su, R., Wang, J., Wang, F., Cai, L., Zhao, Y., et al. (2022) Observation of Five Types of Leaders Contained in a Negative Triggered Lightning. Scientific Reports, 12, Article No. 6299. [Google Scholar] [CrossRef] [PubMed]
[19] Cai, L., Chu, W., Zhou, M., Su, R., Xu, C., Wang, J., et al. (2023) Observation and Modeling of Attempted Leaders in a Multibranched Altitude-Triggered Lightning Flash. IEEE Transactions on Electromagnetic Compatibility, 65, 1133-1142. [Google Scholar] [CrossRef] 
[20] Winckler, J.R., Malcolm, P.R., Arnoldy, R.L., Burke, W.J., Erickson, K.N., Ernstmeyer, J., et al. (1989) ECHO 7: An Electron Beam Experiment in the Magnetosphere. Eos, Transactions American Geophysical Union, 70, 657-668. [Google Scholar] [CrossRef] 
[21] Pasko, V.P., Stanley, M.A., Mathews, J.D., Inan, U.S. and Wood, T.G. (2002) Electrical Discharge from a Thundercloud Top to the Lower Ionosphere. Nature, 416, 152-154. [Google Scholar] [CrossRef] [PubMed]
[22] Soula, S., van der Velde, O., Montanyà, J., Neubert, T., Chanrion, O. and Ganot, M. (2009) Analysis of Thunderstorm and Lightning Activity Associated with Sprites Observed during the Eurosprite Campaigns: Two Case Studies. Atmospheric Research, 91, 514-528. [Google Scholar] [CrossRef] 
[23] Yang, J., Lu, G., Liu, N., Cui, H., Wang, Y. and Cohen, M. (2017) Analysis of a Mesoscale Convective System That Produced a Single Sprite. Advances in Atmospheric Sciences, 34, 258-271. [Google Scholar] [CrossRef] 
[24] Wang, X., Wang, H., Lyu, W., Chen, L., Ma, Y., Qi, Q., et al. (2022) First Experimental Verification of Opacity for the Lightning Near‐Infrared Spectrum. Geophysical Research Letters, 49, GL098883. [Google Scholar] [CrossRef] 
[25] Wan, R., Yuan, P., An, T., Liu, G., Wang, X., Wang, W., et al. (2021) Effects of Atmospheric Attenuation on the Lightning Spectrum. Journal of Geophysical Research: Atmospheres, 126, JD035387. [Google Scholar] [CrossRef] 
[26] Lecun, Y., Bottou, L., Bengio, Y. and Haffner, P. (1998) Gradient-Based Learning Applied to Document Recognition. Proceedings of the IEEE, 86, 2278-2324. [Google Scholar] [CrossRef] 
[27] Krizhevsky, A., Sutskever, I. and Hinton, G.E. (2012) ImageNet Classification with Deep Convolutional Neural Networks. Proceedings of the 25th International Conference on Neural Information Processing Systems, Lake Tahoe, 3-6 December 2012, 1097-1105.
[28] Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., et al. (2015) Going Deeper with Convolutions. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, 7-12 June 2015, 1-9. [Google Scholar] [CrossRef] 
[29] He, K., Zhang, X., Ren, S. and Sun, J. (2016) Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, 27-30 June 2016, 770-778. [Google Scholar] [CrossRef] 
[30] Zagoruyko, S. and Komodakis, N. (2016) Wide Residual Networks. arXiv: 1605.07146. [Google Scholar] [CrossRef] 
[31] Xie, S., Girshick, R., Dollar, P., Tu, Z. and He, K. (2017) Aggregated Residual Transformations for Deep Neural Networks. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, 21-26 July 2017, 5987-5995. [Google Scholar] [CrossRef] 
[32] Huang, G., Liu, Z., Van Der Maaten, L. and Weinberger, K.Q. (2017) Densely Connected Convolutional Networks. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, 21-26 July 2017, 2261-2269. [Google Scholar] [CrossRef] 
[33] Zhao, B., Li, X., Lu, X. and Wang, Z. (2018) A CNN-RNN Architecture for Multi-Label Weather Recognition. Neurocomputing, 322, 47-57. [Google Scholar] [CrossRef] 
[34] Lv, Q., Li, Q., Chen, K., Lu, Y. and Wang, L. (2022) Classification of Ground-Based Cloud Images by Contrastive Self-Supervised Learning. Remote Sensing, 14, Article 5821. [Google Scholar] [CrossRef] 
[35] Kohlmann, H., Schulz, W. and Pedeboy, S. (2017) Evaluation of EUCLID IC/CG Classification Performance Based on Ground-Truth Data. 2017 International Symposium on Lightning Protection (XIV SIPDA), Natal, 2-6 October 2017, 35-41. [Google Scholar] [CrossRef] 
[36] Nag, A., Murphy, M.J., Cummins, K.L., et al. (2014) Recent Evolution of the Us National Lightning Detection Network. 23rd International Lightning Detection Conference & 5th International Lightning Meteorology Conference, Tucson, 18-21 March 2014, 1-6.
[37] Zhou, K., Zheng, Y., Dong, W. and Wang, T. (2020) A Deep Learning Network for Cloud-to-Ground Lightning Nowcasting with Multisource Data. Journal of Atmospheric and Oceanic Technology, 37, 927-942. [Google Scholar] [CrossRef] 
[38] 段瑞玲, 李庆祥, 李玉和. 图像边缘检测方法研究综述[J]. 光学技术, 2005, 31(3): 415-419.
[39] 马艳, 张治辉. 几种边缘检测算子的比较[J]. 工矿自动化, 2004, 30(1): 54-56.

为你推荐