基于DnNet的探地雷达地下管线数据降噪方法

doi:10.11947/j.AGCS.2025.20230508

测绘学报 ›› 2025, Vol. 54 ›› Issue (8): 1464-1475.doi: 10.11947/j.AGCS.2025.20230508

基于DnNet的探地雷达地下管线数据降噪方法

王惠琴¹(), 李佳豪¹, 刘鑫¹, 何永强², 罗佳¹, 刘宾灿³

^1.兰州理工大学计算机与通信学院，甘肃　兰州　730050
^2.西北民族大学土木工程学院，甘肃　兰州　730030
^3.陕西建工安装集团有限公司，陕西　西安　710068

收稿日期:2024-01-01 修回日期:2025-07-08 出版日期:2025-09-16 发布日期:2025-09-16
作者简介:王惠琴（1971—），女，博士，教授，研究方向为雷达信号检测与处理。E-mail：whq1222@lut.edu.cn
基金资助:
甘肃省重点研发计划(23YFFA0060);甘肃省优秀研究生“创新之星”项目(2025CXZX-57)

A denoising method for underground pipeline data acquired by ground penetrating radar based on DnNet

Huiqin WANG¹(), Jiahao LI¹, Xin LIU¹, Yongqiang HE², Jia LUO¹, Bincan LIU³

^1.School of Computer and Communication, Lanzhou University of Technology, Lanzhou 730050, China
^2.School of Civil Engineering, Northwest Minzu University, Lanzhou 730030, China
^3.SCEGC Equipment Installation Group Co., Ltd., Xi'an 710068, China

Received:2024-01-01 Revised:2025-07-08 Online:2025-09-16 Published:2025-09-16
About author:WANG Huiqin (1971—), female, PhD, professor, majors in radar signal detection and processing. E-mail: whq1222@lut.edu.cn
Supported by:
Key Research and Development Program of Gansu Province(23YFFA0060);“Innovation Star” Project of Outstanding Graduate Students in Gansu Province(2025CXZX-57)

摘要/Abstract

摘要：

噪声的存在会严重影响探地雷达（ground-penetrating radar，GPR）地下管线的智能解译和识别，鉴于此，本文提出一种基于DnNet的探地雷达地下管线数据降噪方法。其中，该算法利用编码器-解码器结构、组归一化和简化的通道注意力机制构造了全新的深度学习降噪网络，实现了探地雷达图像降噪性能的大幅提升。利用深度卷积块改进前馈网络，有效提高了网络对波形边缘信息的恢复能力。同时，也因简化通道注意力机制和前馈网络的改进，大幅度提高了降噪效率。试验结果表明，本文算法有良好的降噪效果。在模拟GPR图像降噪中，相较于字典学习方法、Cycle GAN、DRUNet和DnCNN，当噪声标准差等于50时，本文算法的峰值信噪比分别提升了24.72、24.3、23.54和23.86 dB，结构相似性分别提升了0.545 5、0.424 2、0.140 8和0.375 9。在实际GPR数据降噪中，本文算法相较其他算法能够去除大部分噪声并保留地下管线的波形细节。

关键词: GPR图像降噪, 编码器-解码器, 组归一化, 通道注意力

Abstract:

The presence of noise can seriously affect the intelligent interpretation and identification of ground-penetrating radar (GPR) underground pipelines. In view of this, this paper proposes a PnNet-based denoising method for GPR under ground pipeline data. Among them, this algorithm constructs a brand-new deep learning denoising network by using the encoder-decoder structure, group normalization and simplified channel attention mechanism, achieving a significant improvement in the denoising performance of ground penetrating radar images. The feedforward network is improved by using deep convolutional blocks, effectively enhancing the network's ability to recover waveform edge information. Meanwhile, due to the simplification of the channel attention mechanism and the improvement of the feedforward network, the noise reduction efficiency has been significantly enhanced. The experimental results show that the proposed algorithm has a good noise reduction effect. In the noise reduction of simulated GPR images, compared with the dictionary learning method, Cycle GAN, DRUNet and DnCNN, when the standard deviation of noise is equal to 50, the peak signal-to-noise ratio of the proposed algorithm has increased by 24.72, 24.3, 23.54 and 23.86 dB respectively. The structural similarities have increased by 0.545 5, 0.424 2, 0.140 8 and 0.375 9 respectively. In the actual denoising of GPR data, the proposed algorithm can remove most of the noise and retain the waveform details of underground pipelines compared with other algorithms.

Key words: GPR image denoising, encoder-decoder, group normalization, channel attention

中图分类号:

P237

王惠琴, 李佳豪, 刘鑫, 何永强, 罗佳, 刘宾灿. 基于DnNet的探地雷达地下管线数据降噪方法[J]. 测绘学报, 2025, 54(8): 1464-1475.

Huiqin WANG, Jiahao LI, Xin LIU, Yongqiang HE, Jia LUO, Bincan LIU. A denoising method for underground pipeline data acquired by ground penetrating radar based on DnNet[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(8): 1464-1475.

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: http://xb.chinasmp.com/CN/10.11947/j.AGCS.2025.20230508

http://xb.chinasmp.com/CN/Y2025/V54/I8/1464

图/表 10

图1

图2

图3

表1

图4

图5

图6

表2

图7

图8

参考文献 28

[1]	LI Jinghe, HE Zhanxiang, YANG Jun, et al. Scale and rotation statistic-based self-adaptive function for ground penetrating radar denoising in curvelet domain[J]. Acta Physica Sinica, 2019, 68(9): 090501.
[2]	杨必胜, 宗泽亮, 陈驰, 等. 车载探地雷达地下目标实时探测法[J]. 测绘学报, 2020, 49(7): 874-882. DOI: . doi: 10.11947/j.AGCS.2020.20190293
	YANG Bisheng, ZONG Zeliang, CHEN Chi, et al. Real time approach for underground objects detection from vehicle-borne ground penetrating radar[J]. Acta Geodaetica et Cartographica Sinica, 2020, 49(7): 874-882. DOI: . doi: 10.11947/j.AGCS.2020.20190293
[3]	王大为, 王召巴. 一种强噪声背景下微弱超声信号提取方法研究[J]. 物理学报, 2018, 67(21): 65-77.
	WANG Dawei, WANG Zhaoba. Weak ultrasonic signal detection in strong noise[J]. Acta Physica Sinica, 2018, 67(21): 65-77.
[4]	张立国, 周正欧. 浅地层探地雷达回波倒相的自适应处理[J]. 电子科技大学学报, 2004, 33(5): 519-522.
	ZHANG Liguo, ZHOU Zheng'ou. Selfadapting processing of the reflecting signal phase invertion of subsurface ground penetrating radar[J]. Journal of University of Electronic Science and Technology of China, 2004, 33(5): 519-522.
[5]	黄长军, 郭际明, 喻小东, 等. 干涉图EMD-自适应滤波去噪法[J]. 测绘学报, 2013, 42(5): 707-714.
	HUANG Changjun, GUO Jiming, YU Xiaodong, et al. The study of interferogram denoising method based on EMD and adaptive filter[J]. Acta Geodaetica et Cartographica Sinica, 2013, 42(5): 707-714.
[6]	ZOU Hailin, YANG Feng. Image denoising of ground penetrating radar based on wavelet scale space correlation[C]//Proceedings of the 1st International Workshop on Education Technology and Computer Science. Wuhan: IEEE, 2009: 499-503.
[7]	王昶, 张永生, 王旭, 等. 遥感影像条带噪声去除的小波变分法[J]. 测绘学报, 2019, 48(8): 1025-1037. DOI: . doi: 10.11947/j.AGCS.2019.20180394
	WANG Chang, ZHANG Yongsheng, WANG Xu, et al. Stripe noise removal of remote image based on wavelet variational method[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(8): 1025-1037. DOI: . doi: 10.11947/j.AGCS.2019.20180394
[8]	WANG Chang, ZHANG Yongsheng, WANG Xu, et al. An effective strip noise removal method for remote sensing image[J]. Journal of Geodesy and Geoinformation Science, 2022, 5(4): 72-85.
[9]	QU Xiaofei, ZHAO Weiwei, LONG En, et al. Removal of stripes in remote sensing images based on statistics combined with image enhancement[J]. Journal of Geodesy and Geoinformation Science, 2023, 6(1): 76-87.
[10]	XUE Shuqiang, YANG Yuanxi. Adjustment model and colored noise compensation of continuous observation system[J]. Journal of Geodesy and Geoinformation Science, 2018, 1(1): 39-45.
[11]	武曙光, 边少锋, 李厚朴, 等. 基于变分模态分解的GNSS高程时间序列时变信号提取[J]. 测绘学报, 2024, 53(1): 79-90. DOI: . doi: 10.11947/j.AGCS.2024.20220673
	WU Shuguang, BIAN Shaofeng, LI Houpu, et al. Extraction of time-varying signals from GNSS height time series by variational mode decomposition[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(1): 79-90. DOI: . doi: 10.11947/j.AGCS.2024.20220673
[12]	TEMLIOGLU E, ERER I. A novel convolutional autoencoder-based clutter removal method for buried threat detection in ground-penetrating radar[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 3098122.
[13]	YAN Qiuyu, ZHAO Wufan, HUANG Xiao, et al. Automated delineation of smallholder farm fields using fully convolutional networks and generative adversarial networks[J]. Journal of Geodesy and Geoinformation Science, 2022, 5(4): 10-22.
[14]	LIU Lei, CAO Ligang, LU Congde, et al. A denoising method based on cyclegan with attention mechanisms for improving the hidden distress features of pavement[J]. Scientific Reports, 2023, 13(1): 13910.
[15]	林皓, 肖建平, 刘志航, 等. 基于深度学习的铁路路基雷达检测信号中强干扰压制方法研究[J]. 地球物理学进展, 2023, 38(6): 2714-2723.
	LIN Hao, XIAO Jianping, LIU Zhihang, et al. Clutters suppression in GPR signal for railway subgrade detection based on deep learning[J]. Progress in Geophysics, 2023, 38(6): 2714-2723.
[16]	ZHANG Kai, ZUO Wangmeng, CHEN Yunjin, et al. Beyond a Gaussian denoiser: residual learning of deep CNN for image denoising[J]. IEEE Transactions on Image Processing, 2017, 26(7): 3142-3155.
[17]	ZHANG Kai, LI Yawei, ZUO Wangmeng, et al. Plug-and-play image restoration with deep denoiser prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(10): 6360-6376.
[18]	WU Y, HE K. Group normalization[C]//Proceedings of 2018 European Conference on Computer Vision. [S.l.]: IEEE, 2018: 3-19.
[19]	DONG Linhao, XU Shuang, XU Bo. Speech-Transformer: a no-recurrence sequence-to-sequence model for speech recognition[C]//Proceedings of 2018 IEEE International Conference on Acoustics, Speech and Signal Processing. Calgary: IEEE, 2018: 5884-5888.
[20]	ZAMIR S W, ARORA A, KHAN S, et al. Restormer: efficient transformer for high-resolution image restoration[C]//Proceedings of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022: 5718-5729.
[21]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[J]. Advances in Neural Information Processing Systems, 2017, 1: e30.
[22]	ZHANG Yulun, LI Kunpeng, LI Kai, et al. Image super-resolution using very deep residual channel attention networks[C]//Proceedings of 2018 European conference on computer vision. Cham: Springer, 2018: 294-310.
[23]	NAIR V, HINTON G E. Rectified linear units improve restricted boltzmann machines[C]//Proceedings of the 27th International Conference on Machine Learning. [S.l.]: IEEE, 2010.
[24]	YIN Xinyou, GOUDRIAAN J, LANTINGA E A, et al. A flexible sigmoid function of determinate growth[J]. Annals of Botany, 2003, 91(3): 361-371.
[25]	NI Sen, JIA Pengfei, XU Yang, et al. Prediction of CO concentration in different conditions based on Gaussian-TCN[J]. Sensors and Actuators B: Chemical, 2023, 376: 133010.
[26]	CHEN L, CHU X, ZHANG X, et al. Simple baselines for image restoration[C]//Proceedings of the 17th European Conference. Cham: Springer Nature Switzerland, 2022: 17-33.
[27]	WU Haiping, XIAO Bin, CODELLA N, et al. CvT: introducing convolutions to vision transformers[C]//Proceedings of 2021 IEEE/CVF International Conference on Computer Vision. Montreal: IEEE, 2021: 22-31.
[28]	ZHANG Fengzhen, CEN Yigang, ZHAO Ruizhen, et al. Analytic separable dictionary learning based on oblique manifold[J]. Neurocomputing, 2017, 236: 32-38.

深度模型	Depth	Width	Small	Middle	Large
编码器深度	[2，2，4，8]	[2，2，2，2]	[2，2，2，2]	[2，2，4，8]	[4，6，6，8]
瓶颈层深度	12	24	12	24	24
解码器深度	[2，2，2，2]	[2，2，2，2]	[2，2，2，2]	[2，2，2，2]	[2，2，2，2]
训练时间/h	11.5	17.8	13.5	20.9	25.1

深度学习算法	DRUNet	DnCNN	DnNet
计算复杂度MACs	2.0×10¹⁰	1.7×10¹⁰	2.2×10¹⁰
推断速度/s	1.53	0.79	1.74

基于DnNet的探地雷达地下管线数据降噪方法

A denoising method for underground pipeline data acquired by ground penetrating radar based on DnNet

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 28

相关文章 15

编辑推荐

Metrics

本文评价

[1]	李鹏, 张家涵, 汪志翰, 王厚杰, 李振洪. 潮间带地形重建方法综述：现状、挑战与趋势[J]. 测绘学报, 2026, 55(4): 571-587.
[2]	付波霖, 黄柯越, 杨艳丽, 孙伟伟, 王朝茵. 基于实测全谱段高光谱数据的多场景红树林土壤光谱响应特性解析及土壤有机碳含量反演[J]. 测绘学报, 2026, 55(4): 604-617.
[3]	吴岚昕, 彭江涛, 孙伟伟, 杨冰. 面向海岸带湿地高光谱遥感的欧拉映射与互补特征建模变化检测方法[J]. 测绘学报, 2026, 55(4): 618-631.
[4]	高二涛, 刘静, 李淑瑾, 周国清, 付波霖, 李淑娴. 光学遥感与SAR协同的茅尾海潮滩多维时空演变监测与分析[J]. 测绘学报, 2026, 55(4): 632-646.
[5]	徐豪, 徐南, 辛会超, 马跃, 涂伟, 李清泉. 基于ICESat-2卫星高度计数据的潮间带地形光子提取方法[J]. 测绘学报, 2026, 55(4): 658-672.
[6]	布金伟, 刘淑慧, 徐顺双, 向彤粟, 汪秋兰, 籍超颖, 左小清. 星载GNSS-R全球海浪波周期估计的经验模型构建[J]. 测绘学报, 2026, 55(4): 684-697.
[7]	杨泽鑫. 基于点云数据的建筑物多细节层次模型重建关键技术研究[J]. 测绘学报, 2026, 55(4): 755-755.
[8]	姚永祥. 基于广义相似性特征的多模态遥感影像稳健匹配[J]. 测绘学报, 2026, 55(4): 757-757.
[9]	续东. 基于多源三维底座数据的全局定位关键技术研究[J]. 测绘学报, 2026, 55(4): 760-760.
[10]	叶晨鸣, 康志忠, 才谨豪, 左秉正, 邵帅, 李彦. 形态特征引导的月表地理实体实景三维建模方法[J]. 测绘学报, 2026, 55(3): 525-535.
[11]	王美莲. 基于手持激光雷达点云的树种分类及枝叶分离研究[J]. 测绘学报, 2026, 55(3): 569-569.
[12]	彭代锋, 刘雪莲, 鲁梦飞, 管海燕. 基于多尺度跨模态特征融合的异源遥感影像洪水变化检测[J]. 测绘学报, 2026, 55(2): 328-343.
[13]	胡浩鹏, 吴杭彬, 战仕浩, 温在豪, 刘春. 视觉点云质量优化支持的道路杆状物变化检测[J]. 测绘学报, 2026, 55(2): 344-358.
[14]	余东行. 高分辨率遥感影像场景与目标识别技术研究[J]. 测绘学报, 2026, 55(2): 377-377.
[15]	韩斌, 黄欣, 李丰毅, 卢晓珍. 一种双编码器自适应特征融合的SAR图像水体分割网络[J]. 测绘学报, 2026, 55(1): 101-113.