基于U-Net和跨特征图注意力机制的高压电线分割方法
A High-Voltage Power Line Segmentation Method Based on U-Net with Across-Feature Graph Attention Mechanism
摘要: 高压电线巡检对电网维护至关重要，它直接关系到电力系统的安全和人们的日常生活。随着机器学习和深度学习技术的不断发展，图像中高压电线的分割成为了智能巡检领域的研究热点。然而，由于背景复杂、对比度低以及高压电线在图像中占比小等因素，传统的高压电线分割方法往往难以达到理想的精度。为了解决这一问题，本研究在U-Net网络中引入了跨特征图注意力(Across-Feature Map Attention)机制，以增强网络对细小目标特征的学习能力。实验结果表明，在U-Net的第二层深度处加入该注意力机制后，与原始U-Net相比，准确率提高了6.35%，召回率提高了14.84%，F1分数提高了11.05%。
Abstract: High-voltage power line inspection is crucial for grid maintenance, which is directly related to the safety of the power system and people’s daily lives. With the advancement of machine learning and deep learning technology, the segmentation of high-voltage power lines in images has become one of the research hotspots in intelligent inspection. However, due to factors such as complex background, low contrast, and the small proportion of high-voltage power lines in images, traditional high-voltage power line segmentation methods often make it difficult to achieve the desired accuracy. To solve this, this study introduces the Across-Feature Map Attention mechanism into the U-Net network to enhance the network’s ability to learn fine target features. The experimental results show that the U-Net with this attention mechanism added at the second layer depth outperforms the original U-Net, showing a 6.35% improvement in accuracy, a 14.84% increase in recall, and an 11.05% rise in the F1 score, respectively.
文章引用：陈梦洁, 洪智勇, 熊利平. 基于U-Net和跨特征图注意力机制的高压电线分割方法[J]. 图像与信号处理, 2025, 14(2): 199-212. https://doi.org/10.12677/jisp.2025.142019

参考文献

[1] Nguyen, V.N., Jenssen, R. and Roverso, D. (2018) Automatic Autonomous Vision-Based Power Line Inspection: A Review of Current Status and the Potential Role of Deep Learning. International Journal of Electrical Power & Energy Systems, 99, 107-120. [Google Scholar] [CrossRef] 
[2] Ronneberger, O., Fischer, P. and Brox, T. (2015) U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Lecture Notes in Computer Science, Springer, 234-241. [Google Scholar] [CrossRef] 
[3] Abdelfattah, R., Wang, X. and Wang, S. (2021) TTPLA: An Aerial-Image Dataset for Detection and Segmentation of Transmission Towers and Power Lines. In: Lecture Notes in Computer Science, Springer, 601-618. [Google Scholar] [CrossRef] 
[4] Sang, S., Zhou, Y., Islam, M.T. and Xing, L. (2023) Small-Object Sensitive Segmentation Using across Feature Map Attention. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45, 6289-6306. [Google Scholar] [CrossRef] [PubMed]
[5] Du, S., van Wyk, B.J. and Tu, C. (2010) Heuristic Bayesian Pixel Classification for Power Line Inspection. 2010 3rd International Congress on Image and Signal Processing, Yantai, 16-18 October 2010, 960-963. [Google Scholar] [CrossRef] 
[6] Li, Z., Liu, Y., Hayward, R., Zhang, J. and Cai, J. (2008) Knowledge-Based Power Line Detection for UAV Surveillance and Inspection Systems. 2008 23rd International Conference Image and Vision Computing New Zealand, Christchurch, 26-28 November 2008, 1-6. [Google Scholar] [CrossRef] 
[7] Song, B. and Li, X. (2014) Power Line Detection from Optical Images. Neurocomputing, 129, 350-361. [Google Scholar] [CrossRef] 
[8] Ceron, A., Mondragon B., I.F. and Prieto, F. (2014) Power Line Detection Using a Circle Based Search with UAV Images. 2014 International Conference on Unmanned Aircraft Systems (ICUAS), Orlando, 27-30 May 2014, 632-639. [Google Scholar] [CrossRef] 
[9] Archana, R. and Jeevaraj, P.S.E. (2024) Deep Learning Models for Digital Image Processing: A Review. Artificial Intelligence Review, 57, Article No. 11. [Google Scholar] [CrossRef] 
[10] Ahmed, S.A., Desa, H., Easa, H.K., Hussain, A.T., Taha, T.A., Salih, S.Q., et al. (2024) Advancements in UAV Image Semantic Segmentation: A Comprehensive Literature Review. Multidisciplinary Reviews, 7, Article 2024118. [Google Scholar] [CrossRef] 
[11] Xu, G., Wang, X., Wu, X., et al. (2024) Development of Skip Connection in Deep Neural Networks for Computer Vision and Medical Image Analysis: A Survey.
[12] Iqra Giri, K.J. and Javed, M. (2024) Small Object Detection in Diverse Application Landscapes: A Survey. Multimedia Tools and Applications, 83, 88645-88680. [Google Scholar] [CrossRef] 
[13] Wang, H. and Gao, P. (2024) Survey of Small Object Detection Methods Based on Deep Learning. 2024 9th International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS), Okinawa, 21-23 November 2024, 221-224. [Google Scholar] [CrossRef] 
[14] Shelhamer, E., Long, J. and Darrell, T. (2014) Fully Convolutional Networks for Semantic Segmentation. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, 7-12 June 2014, 3431-3440.
[15] Badrinarayanan, V., Kendall, A. and Cipolla, R. (2017) SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39, 2481-2495. [Google Scholar] [CrossRef] [PubMed]
[16] Zhao, H., Shi, J., Qi, X., Wang, X. and Jia, J. (2017) Pyramid Scene Parsing Network. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, 21-26 July 2017, 6230-6239. [Google Scholar] [CrossRef] 
[17] Chen, L.-C., Papandreou, G., Kokkinos, I., et al. (2014) Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs.
[18] Chen, L., Papandreou, G., Kokkinos, I., Murphy, K. and Yuille, A.L. (2018) DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40, 834-848. [Google Scholar] [CrossRef] [PubMed]
[19] Yu, F. and Koltun, V. (2015) Multi-Scale Context Aggregation by Dilated Convolutions.
[20] Chen, L.-C., Papandreou, G., Schroff, F., et al. (2017) Rethinking Atrous Convolution for Semantic Image Segmentation.
[21] Lin, G., Milan, A., Shen, C. and Reid, I. (2017) RefineNet: Multi-Path Refinement Networks for High-Resolution Semantic Segmentation. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, 21-26 July 2017, 5168-5177. [Google Scholar] [CrossRef] 
[22] Madaan, R., Maturana, D. and Scherer, S. (2017) Wire Detection Using Synthetic Data and Dilated Convolutional Networks for Unmanned Aerial Vehicles. 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Vancouver, 24-28 September 2017, 3487-3494. [Google Scholar] [CrossRef] 
[23] Wang, X. (2019) Research on Semantic Segmentation of Power Line Based on Image.
https://kns.cnki.net/kcms2/article/abstract?v=uQzRnDzoTXG67ksS9rxPWD9fh-SqHMF3mS4LID1_KDzsWXtt9ghKv_0Z0uZP5TEaBJSfUX37d1KX9HyA7olN6M0emUkRLkfwjtDcwsOrz6sPN4rdftUKHbQxdo_UWziWRknSBDnej9aapELDZYjJnqxCeuWPjA_RplZkVRo6Tc-aeis-ZiCIXtiG7Pg_w_xBhjvDlLiCkOo=&uniplatform=NZKPT&language=CHS
[24] Zhang, H., Yang, W., Yu, H., Zhang, H. and Xia, G. (2019) Detecting Power Lines in UAV Images with Convolutional Features and Structured Constraints. Remote Sensing, 11, Article 1342. [Google Scholar] [CrossRef] 
[25] Dai, Z., Yi, J., Zhang, Y., Zhou, B. and He, L. (2020) Fast and Accurate Cable Detection Using CNN. Applied Intelligence, 50, 4688-4707. [Google Scholar] [CrossRef] 
[26] Choi, H., Koo, G., Kim, B.J. and Kim, S.W. (2021) Weakly Supervised Power Line Detection Algorithm Using a Recursive Noisy Label Update with Refined Broken Line Segments. Expert Systems with Applications, 165, Article 113895. [Google Scholar] [CrossRef] 
[27] Brown, R.G. (1960) Statistical Forecasting for Inventory Control. McGraw-Hill.
[28] Hu, J., Shen, L. and Sun, G. (2018) Squeeze-and-Excitation Networks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 18-23 June 2018, 7132-7141. [Google Scholar] [CrossRef] 
[29] Woo, S., Park, J., Lee, J. and Kweon, I.S. (2018) CBAM: Convolutional Block Attention Module. In: Lecture Notes in Computer Science, Springer, 3-19. [Google Scholar] [CrossRef] 
[30] Oktay, O., Schlemper, J., Folgoc, L.L., et al. (2018) Attention U-Net: Learning Where to Look for the Pancreas.