Three-dimensional target inversion algorithm based on multi-feature reconstruction

doi:10.3785/j.issn.1008-973X.2024.11.001

Journal of ZheJiang University (Engineering Science)

2024, Vol. 58

Issue (11): 2199-2207 DOI: 10.3785/j.issn.1008-973X.2024.11.001

Three-dimensional target inversion algorithm based on multi-feature reconstruction

Yali XUE1(

),Lizun ZHOU1,Linfei WANG2,Quan OUYANG1

1. College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
2. China Aero Geophysical Survey and Remote Sensing Center for Natural Resources, Beijing 100000, China

Download:

HTML

PDF(2281KB) HTML
Export: BibTeX | EndNote (RIS)

Abstract

A 3D target inversion algorithm based on multi-feature reconstruction was proposed in order to solve the problems of large memory occupation and time-consuming training in deep learning-based three-dimensional inversion methods. Four types of features, horizontal area, center depth, vertical thickness and residual density of the target were extracted by feature decomposition to realize the compression of the three-dimensional model and reduce the memory occupation. The multi-feature reconstruction of inversion network (MRNet) was designed to realize the prediction of the four types of target features by different Decoder, and the four types of features were used to reconstruct the three-dimensional model to realize the inversion of the 3D target. The gradient union was introduced at the input of the network to realize the enhancement of target boundary information. The CA attention mechanism was introduced at the cross-layer connection to realize the differentiation of Decoder’s prediction function and optimize the inversion effect. The simulation results showed that the local relative accuracy of MRNet was improved by more than 30% compared with 3D U-Net, reaching 88.91%, and the training time per round was only 1/13 of 3D U-Net. MRNet was applied to Vinton Salt Mound, and the distribution of caprocks was obtained more accurately, which verified that MRNet had certain generalizability.

Key words： three-dimensional target inversion multi-feature reconstruction attention mechanism deep learning multitask learning

Received: 16 November 2023 Published: 23 October 2024

CLC:	TP 39
	P 31

Fund: 国家自然基金资助项目(62073164)；上海航天科技创新基金资助项目(SAST2022-013).

	Service
	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Yali XUE
	Lizun ZHOU
	Linfei WANG
	Quan OUYANG

Cite this article:

Yali XUE,Lizun ZHOU,Linfei WANG,Quan OUYANG. Three-dimensional target inversion algorithm based on multi-feature reconstruction. Journal of ZheJiang University (Engineering Science), 2024, 58(11): 2199-2207.

URL:

https://www.zjujournals.com/eng/10.3785/j.issn.1008-973X.2024.11.001 OR https://www.zjujournals.com/eng/Y2024/V58/I11/2199

基于多特征重构的三维目标反演算法

为了解决基于深度学习的三维反演方法中存在的内存占用大、训练耗时久的问题，提出基于多特征重构的三维目标反演算法. 通过特征分解提取目标的水平区域、中心深度、垂直厚度和剩余密度4类特征，实现对三维模型的压缩，降低内存占用. 设计多特征重构反演网络(MRNet)，通过不同的Decoder实现对目标4类特征的预测，利用4类特征重构三维模型，实现对三维目标的反演. 在网络输入端引入梯度联合实现对目标边界信息的增强. 在跨层连接处引入CA注意力机制，实现对Decoder预测功能的分化，优化反演效果. 模拟实验结果显示，MRNet的局部相对准确度相对于3D U-Net提升了30%以上，达到88.91%，每轮训练时间仅为3D U-Net的1/13. 将MRNet应用于Vinton盐丘地区，较准确地得到了盖岩的分布情况，验证了MRNet具备一定的泛化性.

关键词： 三维目标反演, 多特征重构, 注意力机制, 深度学习, 多任务学习

Fig.1 Diagram of feature decomposition

Fig.2 Flowchart of gradient data fusion

Fig.3 MRNet network structure

Fig.4 Structural details of encoder-decoder

Fig.5 Structure diagram of coordinate attention module

Fig.6 Diagram of partial simulated anomalous body

Tab.1 Dimensional parameters of anomalies in all directions

Fig.7 MRNet loss function decline curve

Fig.8 Gravity anomaly of test sample

Fig.9 Simulation model of test sample and inverse model of MRNet

Tab.2 Maximum absolute error in test sample feature predictions

Fig.10 Inverse model of 3D U-Net and DecNet

Tab.3 Mean absolute error, global absolute accuracy, and local relative accuracy of three inversion networks

Fig.11 Projection result for three error models of sample D

Fig.12 Gravity data from Vinton salt dome and MRNet inversion result

Fig.13 Triaxial projection of Vinton salt dome inversion result

Fig.14 Inverse result of 3D U-Net and DecNet


[1]	ZHDANOV M S. Inverse theory and applications in geophysics [M]. Netherlands: Elsevier, 2015.

[2]	LAST B J, KUBIK K Compact gravity inversion[J]. Geophysics, 1983, 48 (6): 713- 721 doi: 10.1190/1.1441501

[3]	LI Y, OLDENBURG D W 3-D inversion of gravity data[J]. Geophysics, 1998, 63 (1): 109- 119 doi: 10.1190/1.1444302

[4]	PORTNIAGUINE O, ZHDANOV M S Focusing geophysical inversion images[J]. Geophysics, 1999, 64 (3): 874- 887 doi: 10.1190/1.1444596

[5]	师学明, 王家映, 张胜业, 等多尺度逐次逼近遗传算法反演大地电磁资料[J]. 地球物理学报, 2000, 43 (1): 122- 130 SHI Xueming, WANG Jiaying, ZHANG Shengye, et al Multiscale genetic algorithm and its application in magnetotelluric sounding data inversion[J]. Chinese Journal of Geophysics, 2000, 43 (1): 122- 130

[6]	罗红明, 王家映, 朱培民, 等量子遗传算法在大地电磁反演中的应用[J]. 地球物理学报, 2009, 52 (1): 260- 267 LUO Hongming, WANG Jiaying, ZHU Peimin, et al Quantum genetic algorithm and its application in magnetotelluric data inversion[J]. Chinese Journal of Geophysics, 2009, 52 (1): 260- 267 doi: 10.1002/cjg2.1347

[7]	陈华根, 李嘉虓, 吴健生, 等 MT-重力模拟退火联合反演研究[J]. 地球物理学报, 2012, 55 (2): 663- 670 CHEN Huagen, LI Jiaxiao, WU Jiansheng, et al Study on simulated-annealing MT-gravity joint inversion[J]. Chinese Journal of Geophysics, 2012, 55 (2): 663- 670

[8]	邱宁, 刘庆生, 曾佐勋, 等基于混沌-粒子群优化的磁法数据非线性反演方法[J]. 地球物理学进展, 2010, (6): 2150- 2155 QIU Ning, LIU Qingsheng, ZENG Zuoxun, et al Nonlinear inversion of magnetic data based on chaotic and particle swarm optimization[J]. Progress in Geophysics, 2010, (6): 2150- 2155

[9]	刘双, 刘天佑, 冯杰, 等蚁群算法在磁测资料反演解释中的应用[J]. 物探与化探, 2013, 37 (1): 150- 154 LIU Shuang, LIU Tianyou, FENG Jie, et al The application of ant colony algorithm to the inversion and interpretation of magnetic data[J]. Geophysical and Geochemical Exploration, 2013, 37 (1): 150- 154

[10]	王逸宸, 柳林涛, 许厚泽基于卷积神经网络识别重力异常体[J]. 物探与化探, 2020, 44 (2): 394- 400 WANG Yichen, LIU Lintao, XU Houze The identification of gravity anomaly body based on the convolutional neural network[J]. Geophysical and Geochemical Exploration, 2020, 44 (2): 394- 400

[11]	阳前果. 基于深度学习的重磁数据处理研究[D]. 武汉: 中国地质大学, 2021. YANG Qianguo. Research on gravity and magnetic data processing based on deep learning [D]. Wuhan: China University of Geosciences, 2021.

[12]	RONNEBERGER O, FISCHER P, BROX T. U-net: convolutional networks for biomedical image segmentation [C]// Medical Image Computing and Computer-Assisted Intervention: 18th International Conference. Munich: Springer, 2015: 234−241.

[13]	WANG Yufeng, ZHANG Yujie, FU Lihua, et al Three-dimensional gravity inversion based on 3D U-Net++[J]. Applied Geophysics, 2021, 18 (4): 451- 460 doi: 10.1007/s11770-021-0909-z

[14]	ZHANG L, ZHANG G, LIU Y, et al Deep learning for 3-D inversion of gravity data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60: 1- 18

[15]	ZHANG S, YIN C, CAO X, et al DecNet: decomposition network for 3D gravity inversion[J]. Geophysics, 2022, 87 (5): G103- G114 doi: 10.1190/geo2021-0744.1

[16]	RUDER S. An overview of gradient descent optimization algorithms [EB/OL]. (2016-09-01) [2023-10-01]. https://arxiv.org/pdf/1609.04747.pdf.

[17]	HOU Q, ZHOU D, FENG J. Coordinate attention for efficient mobile network design [C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway: IEEE, 2021: 13713-13722.

[18]	KINGMA D P, BA J L. ADAM: a method for stochastic optimization [EB/OL]. (2014-12-01) [2023-10-01]. https://arxiv.org/pdf/1412.6980.pdf.

[19]	ENNEN C. Mapping gas-charged fault blocks around the Vinton Salt Dome [D]. Houston: University of Houston, 2012.

[20]	OLIVEIRA JR V C, BARBOSA V C F 3-D radial gravity gradient inversion[J]. Geophysical Journal International, 2013, 195 (2): 883- 902 doi: 10.1093/gji/ggt307

[21]	高秀鹤, 曾昭发, 孙思源, 等基于阈值约束的协克里金法联合反演重力与重力梯度数据[J]. 地球物理学报, 2019, 62 (3): 1037- 1045 GAO Xiuhe, ZENG Zhaofa, SUN Siyuan, et al Joint inversion of gravity and gravity gradient data based on Cokriging method with the threshold constrain[J]. Chinese Journal of Geophysics, 2019, 62 (3): 1037- 1045

[22]	侯振隆, 王恩德, 周文纳, 等重力梯度欧拉反褶积及其在文顿岩丘的应用[J]. 石油地球物理勘探, 2019, 54 (2): 472- 479 HOU Zhenlong, WANG Ende, ZHOU Wenna, et al Gravity gradient Euler deconvolution and its application to Vinton Salt Dome[J]. Oil Geophysical Prospecting, 2019, 54 (2): 472- 479

[1]	Fan LI,Jie YANG,Zhicheng FENG,Zhichao CHEN,Yunxiao FU. Pantograph-catenary contact point detection method based on image recognition[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(9): 1801-1810.

[2]	Li XIAO,Zhigang CAO,Haoran LU,Zhijian HUANG,Yuanqiang CAI. Elastic metamaterial design based on deep learning and gradient optimization[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(9): 1892-1901.

[3]	Canlin LI,Xinyue WANG,Lizhuang MA,Zhiwen SHAO,Wenjiao ZHANG. Image cartoonization incorporating attention mechanism and structural line extraction[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(8): 1728-1737.

[4]	Zhongliang LI,Qi CHEN,Lin SHI,Chao YANG,Xianming ZOU. Dynamic knowledge graph completion of temporal aware combination[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(8): 1738-1747.

[5]	Shuhan WU,Dan WANG,Yuanfang CHEN,Ziyu JIA,Yueqi ZHANG,Meng XU. Attention-fused filter bank dual-view graph convolution motor imagery EEG classification[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(7): 1326-1335.

[6]	Linrui LI,Dongsheng WANG,Hongjie FAN. Fact-based similar case retrieval methods based on statutory knowledge[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(7): 1357-1365.

[7]	Xianwei MA,Chaohui FAN,Weizhi NIE,Dong LI,Yiqun ZHU. Robust fault diagnosis method for failure sensors[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(7): 1488-1497.

[8]	Jun YANG,Chen ZHANG. Semantic segmentation of 3D point cloud based on boundary point estimation and sparse convolution neural network[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(6): 1121-1132.

[9]	Yuntang LI,Hengjie LI,Kun ZHANG,Binrui WANG,Shanyue GUAN,Yuan CHEN. Recognition of complex power lines based on novel encoder-decoder network[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(6): 1133-1141.

[10]	Juan SONG,Longxi HE,Huiping LONG. Deep learning-based algorithm for multi defect detection in tunnel lining[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(6): 1161-1173.

[11]	Zhiwei XING,Shujie ZHU,Biao LI. Airline baggage feature perception based on improved graph convolutional neural network[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(5): 941-950.

[12]	Yi LIU,Yidan CHEN,Lin GAO,Jiao HONG. Lightweight road extraction model based on multi-scale feature fusion[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(5): 951-959.

[13]	Cuiting WEI,Weijian ZHAO,Bochao SUN,Yunyi LIU. Intelligent rebar inspection based on improved Mask R-CNN and stereo vision[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(5): 1009-1019.

[14]	Bo ZHONG,Pengfei WANG,Yiqiao WANG,Xiaoling WANG. Survey of deep learning based EEG data analysis technology[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(5): 879-890.

[15]	Hai HUAN,Yu SHENG,Chenxi GU. Global guidance multi-feature fusion network based on remote sensing image road extraction[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(4): 696-707.

Viewed

Full text

Abstract

Cited

Shared

Discussed