基于边缘增强生成对抗网络的电力设备热成像超分辨率重建

doi:10.12204/j.issn.1000-7229.2021.07.010

电力建设 ›› 2021, Vol. 42 ›› Issue (7): 83-89.doi: 10.12204/j.issn.1000-7229.2021.07.010

基于边缘增强生成对抗网络的电力设备热成像超分辨率重建

刘云峰¹, 杨晋彪², 韩晋锋², 彭轶灏³, 赵洪山³

1.国网晋城供电公司,山西省晋城市 048000
2.国网陵川县供电公司,山西省陵川县 048300
3.华北电力大学电力工程系,河北省保定市071003

收稿日期:2020-09-27 出版日期:2021-07-01 发布日期:2021-07-09
通讯作者: 彭轶灏
作者简介:刘云峰(1978),男,学士,高级工程师,主要研究方向为电气工程及其自动化;|杨晋彪(1967),男,学士,高级工程师,主要研究方向为工业电气自动化技术;|韩晋锋(1978),男,学士,高级工程师,主要研究方向为继电保护和信息通信;|赵洪山(1965),男,博士,教授,博士生导师,研究方向为电力系统可靠性、主动配电网和故障预测。
基金资助:
国家自然科学基金青年科学基金项目(51807063)

Super-Resolution Reconstruction of Thermal Imaging of Power Equipment Based on Edge-Enhancement Generative Adversarial Network

LIU Yunfeng¹, YANG Jinbiao², HAN Jinfeng², PENG Yihao³, ZHAO Hongshan³

1. State Grid Jincheng Power Supply Company, Jincheng 048000, Shanxi Province, China
2. State Grid Lingchuan Power Supply Company, Lingchuan 048300, Shanxi Province, China
3. Department of Power Engineering, North China Electric Power University, Baoding 071003, Hebei Province, China

Received:2020-09-27 Online:2021-07-01 Published:2021-07-09
Contact: PENG Yihao
Supported by:
Scientific Funds for Young Scientists of China(51807063)

摘要/Abstract

摘要：

非接触式红外热成像的温度数据,对于电力设备状态监测与健康状态评估具有重要意义。但高分辨率红外成像仪的昂贵造价和技术壁垒,限制了高分辨率热成像图像在基于电力物联网的设备在线监测中的应用。超分辨率重建满足分辨率要求的同时可以降低成本。通过构建改进边缘增强的生成对抗网络实现了电力设备热成像的增强,该网络在超分辨率生成对抗网络(super-resolution generative adversarial networks,SRGAN)的基础上新增了深度残差收缩网络模块,在降低图像噪声的基础上,提高了训练的稳定性,并通过边缘提取技术对图像的峰值信息进行加强重建,提高了边缘恢复效果。利用峰值信噪比(peak signal-to-noise ratio,PSNR)以及结构相似性(structural similarity,SSIM)指标,对重建后的整体数据以及边缘数据进行分析,结果显示,2项指标均显著提高,重建后的主观视觉效果也更加清晰,具有较高的工程实用价值。

关键词: 电力设备热成像, 超分辨率重建, 深度残差收缩网络, 边缘提取

Abstract:

The temperature data of non-contact infrared thermal imaging is of great significance for condition monitoring and health assessment of power equipment. However, the high cost and technical barriers of high-resolution infrared imager limit the application of high-resolution thermal imaging in on-line monitoring of equipment in power internet of things. Super-resolution reconstruction meets the resolution requirements and reduces the cost at the same time. In this paper, the enhancement of thermal imaging of power equipment is realized by constructing an improved generative adversarial network of edge enhancement. The network adds a depth residual shrinkage network module on the basis of SRGAN (super-resolution generative adversarial networks), which improves the stability of training on the basis of reducing image noise, enhances the reconstruction of image peak information through edge-extraction technology, and improves the effect of edge restoration. The example analysis shows that, PSNR (peak signal-to-noise ratio ) and SSIM (structural similarity) indicators are used to analyze the overall data and edge data after reconstruction, both are significantly improved, and the subjective visual effect after reconstruction is more clear, which has high engineering practical value.

Key words: power equipment thermal imaging, super-resolution reconstruction, deep residual shrinkage network, edge extraction

中图分类号:

TM769

刘云峰, 杨晋彪, 韩晋锋, 彭轶灏, 赵洪山. 基于边缘增强生成对抗网络的电力设备热成像超分辨率重建[J]. 电力建设, 2021, 42(7): 83-89.

LIU Yunfeng, YANG Jinbiao, HAN Jinfeng, PENG Yihao, ZHAO Hongshan. Super-Resolution Reconstruction of Thermal Imaging of Power Equipment Based on Edge-Enhancement Generative Adversarial Network[J]. ELECTRIC POWER CONSTRUCTION, 2021, 42(7): 83-89.

图/表 8

图1 EE-SRGAN模型结构原理图

Fig.1 Model structure of EE-SRGAN

图2 残差收缩网络(RSN)

Fig.2 Residual shrinkage network (RSN)

图3 PNSR随迭代次数变化过程

Fig.3 Change process of PSNR with the number of iterations

图4 SSIM随迭代次数变化过程

Fig.4 Change process of SSIM with the number of iterations

表1 不同算法放大4倍重建红外图像评估结果

Table 1 Evaluation of reconstructed infrared image magnified 4 times with different algorithms

表2 不同算法放大4倍重建红外图像边缘数据评估结果

Table 2 Evaluation of edge data of reconstructed infrared image magnified 4 times with different algorithms

图5 重建效果图

Fig.5 Comparison of reconstruction results

图6 重建边缘效果图

Fig.6 Edge comparison of reconstruction results

参考文献 23

[1]	裴少通. 基于红外紫外成像检测技术的绝缘子运行状态分析与评估[D]. 北京: 华北电力大学, 2019.
	PEI Shaotong. Analysis and evaluation of insulator operation status based on infrared and ultraviolet imaging detection technology[D]. Beijing: North China Electric Power University, 2019.
[2]	唐艳秋, 潘泓, 朱亚平, 等. 图像超分辨率重建研究综述[J]. 电子学报, 2020, 48(7):1407-1420.
	TANG Yanqiu, PAN Hong, ZHU Yaping, et al. A survey of image super-resolution reconstruction[J]. Acta Electronica Sinica, 2020, 48(7):1407-1420.
[3]	李伟, 张旭东. 基于卷积神经网络的深度图像超分辨率重建方法[J]. 电子测量与仪器学报, 2017, 31(12):1918-1928.
	LI Wei, ZHANG Xudong. Depth image super-resolution reconstruction based on convolution neural network[J]. Journal of Electronic Measurement and Instrumentation, 2017, 31(12):1918-1928.
[4]	端木春江, 沈碧婷. 基于邻域回归的医学图像超分辨率重建[J]. 计算机应用研究, 2020, 37(12):3792-3794,3802.
	DUANMU Chunjiang, SHEN Biting. Medical image super-resolution reconstruction based on neighborhood regression[J]. Application Research of Computers, 2020, 37(12):3792-3794,3802.
[5]	卢涛, 陈冲, 许若波, 等. 基于边缘增强生成对抗网络的人脸超分辨率重建[J]. 华中科技大学学报(自然科学版), 2020, 48(1):87-92.
	LU Tao, CHEN Chong, XU Ruobo, et al. Face hallucination based on edge enhanced generative adversarial network[J]. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2020, 48(1):87-92.
[6]	JIANG X H, XU Y F, WEI P P, et al. CT image super resolution based on improved SRGAN[C]//2020 5th International Conference on Computer and Communication Systems (ICCCS). Shanghai, China: IEEE, 2020: 363-367.
[7]	赵秀影, 苏耘, 董艳芹, 等. 一种基于小波与双三次插值的CCD图像超分辨方法[J]. 计算机应用研究, 2009, 26(6):2365-2367.
	ZHAO Xiuying, SU Yun, DONG Yanqin, et al. Kind of super-resolution method of CCD image based on wavelet and bicubic interpolation[J]. Application Research of Computers, 2009, 26(6):2365-2367.
[8]	YANG J, WRIGHT J, HUANG T, et al. Image superresolution as sparse representation of raw image patches[C]// 2008 IEEE Conference on Computer Vision and Pattern Recognition. Anchorage, AK, USA:IEEE, 2008.
[9]	DONG C, LOY C C, HE K M, et al. Image super-resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2):295-307. doi: 10.1109/TPAMI.2015.2439281 URL
[10]	DONG C, LOY C C, TANG X O. Accelerating the super-resolution convolutional neural network[M]//Computer Vision - ECCV 2016. Cham: Springer International Publishing, 2016: 391-407.
[11]	SHI W Z, CABALLERO J, HUSZÁR F, et al. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network[J]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2016: 1874-1883.
[12]	KIM J, LEE J K, LEE K M. Accurate image super-resolution using very deep convolutional networks[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA: IEEE, 2016: 1646-1654.
[13]	LEDIG C, THEIS L, HUSZÁR F, et al. Photo-realistic single image super-resolution using a generative adversarial network[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2017: 105-114.
[14]	WANG X T, YU K, WU S X, et al. ESRGAN: Enhanced super-resolution generative adversarial networks[DB/OL]. [2020-04-12]. https://arxiv.org/abs/1809.00219.
[15]	SHANG T Z, DAI Q J, ZHU S C, et al. Perceptual extreme super resolution network with receptive field block[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). IEEE, 2020: 1778-1787.
[16]	CAI J R, ZENG H, YONG H W, et al. Toward real-world single image super-resolution: A new benchmark and a new model[C]//2019 IEEE/CVF International Conference on Computer Vision (ICCV). Seoul, Korea (South): IEEE, 2019: 3086-3095.
[17]	YUAN Y, LIU S Y, ZHANG J W, et al. Unsupervised image super-resolution using cycle-in-cycle generative adversarial networks[C]2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Salt Lake City, UT, USA: IEEE, 2018: 814-81409.
[18]	GUAN J W, PAN C, LI S N, et al. SRDGAN: Learning the noise prior for Super Resolution with Dual Generative Adversarial Networks[DB/OL]. [2020-04-12]..
[19]	ARJOVSKY M, CHINTALA S, BOTTOU L. Wasserstein generative adversarial networks[C]Proceedings of the 34^th International Conference on Machine Learnings . 2017:214-223.
[20]	CHEN Y, TAI Y, LIU X M, et al. FSRNet: end-to-end learning face super-resolution with facial priors[J]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE, 2018: 2492-2501.
[21]	GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial networks[J]. Communications of the ACM, 2020, 63(11):139-144. doi: 10.1145/3422622 URL
[22]	ZHAO M H, ZHONG S S, FU X Y, et al. Deep residual shrinkage networks for fault diagnosis[J]. IEEE Transactions on Industrial Informatics, 2020, 16(7):4681-4690. doi: 10.1109/TII.9424 URL
[23]	Free FLIR thermal dataset for algorithm training[EB/OL]. [2020-06-12]. https://www.flir.com/oem/adas/adas-dataset-form/

基于边缘增强生成对抗网络的电力设备热成像超分辨率重建

Super-Resolution Reconstruction of Thermal Imaging of Power Equipment Based on Edge-Enhancement Generative Adversarial Network

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 23

相关文章 2

编辑推荐

Metrics

本文评价

[1]	李佳玮, 吴克河, 张波. 一种基于小生境遗传算法的电网信息安全风险评估模型[J]. 电力建设, 2021, 42(3): 89-96.
[2]	李福阳, 李俊娥, 刘林彬, 刘威, 林海, 倪明. 智能变电站嵌入式终端安全测试方法研究[J]. 电力建设, 2021, 42(2): 126-136.