DEM Super-Resolution Reconstruction Method Based on Adaptive Generative Adversarial Network

HOU Xujuan; DENG Xiaotong; HUA Weihua; YIN Wanting; LI Peng; QI Chunyu

doi:10.13203/j.whugis20240186

您当前的位置：

首页 >

文章列表页 >

DEM Super-Resolution Reconstruction Method Based on Adaptive Generative Adversarial Network

Photogrammetry and Remote Sensing | 更新时间：2026-11-05

- DEM Super-Resolution Reconstruction Method Based on Adaptive Generative Adversarial Network
- Geomatics and Information Science of Wuhan University Vol. 51, Issue 3, Pages: 557-566(2026)
- 作者机构：
  
  1.中国地质大学（武汉）地理与信息工程学院，湖北武汉，430078
  2.中国地质调查局成都地质调查中心，四川成都，610081
  3.湖北省住房和城乡建设厅智能勘察工程技术创新中心，湖北武汉，430063
  4.中国铁路设计集团有限公司，天津，300308
  5.城市轨道交通数字化建设与测评技术国家工程研究中心，天津，300308
- 作者简介：
- 基金信息：
- DOI：10.13203/j.whugis20240186
  CLC： P223
- Received：31 October 2024，
  
  Published：05 March 2026
- 稿件说明：
移动端阅览
侯旭娟,邓筱曈,花卫华，等.基于自适应生成对抗网络的DEM超分辨率重建方法[J].武汉大学学报（信息科学版）,2026,51(3):557-566.

HOU Xujuan,DENG Xiaotong,HUA Weihua,et al.DEM Super-Resolution Reconstruction Method Based on Adaptive Generative Adversarial Network[J].Geomatics and Information Science of Wuhan University,2026,51(3):557-566.
侯旭娟,邓筱曈,花卫华，等.基于自适应生成对抗网络的DEM超分辨率重建方法[J].武汉大学学报（信息科学版）,2026,51(3):557-566. DOI：10.13203/j.whugis20240186

HOU Xujuan,DENG Xiaotong,HUA Weihua,et al.DEM Super-Resolution Reconstruction Method Based on Adaptive Generative Adversarial Network[J].Geomatics and Information Science of Wuhan University,2026,51(3):557-566. DOI：10.13203/j.whugis20240186

摘要

数字高程模型（digital elevation model

DEM）作为城市建模、灾害评估及三维可视化的数据基础，将其作为输入能够提高许多地理分析应用的准确性和可靠性。为减少重建DEM中的伪影和噪声，提高DEM超分辨率重建的精度，提出了一种结合自适应变形残差卷积与U-Net判别器的超分辨率生成对抗网络（super-resolution generative adversarial network with adaptive deformable residual convolutions and U-Net discriminator

SRDCGAN）方法，实现了DEM的4倍超分辨率重建。对DEM的重建精度进行评估，实验的4组数据集结果表明，均方根误差（root mean square error

RMSE）均值下降了0.3%~12.6%，平均绝对误差（mean absolute error

MAE）均值下降了3.2%~14.5%，坡度误差（slope error

SE）均值下降了1.9%~18.3%，坡向误差（aspect error

AE）均值下降了2.7%~7.1%。将重建效果与双三次插值及超分辨率生成对抗网络方法进行对比，结果显示SRDCGAN方法能够有效去除伪影并保留更多的地形特征，在RMSE、MAE、SE和AE精度评价指标上均有显著改善。

Abstract

Objectives

Digital elevation model (DEM) serves as the fundamental data infrastructure for urban modeling

disaster assessment

and high-fidelity three-dimensional visualization. DEM as an input can significantly enhance the accuracy and reliability of diverse geographic information system applications. To address the inherent limitations of traditional interpolation and standard deep learning models in terrain reconstruction

we aim to develop a super-resolution generative adversarial network with adaptive deformable residual convolutions and U-Net discriminator (SRDCGAN) to suppress reconstruction artifacts and noise while simultaneously elevating the precision of DEM reconstruction.

Methods

The proposed SRDCGAN method features several technical innovations: (1) The feature extraction layer of the generator incorporates a multi-level residual in residual dense block (RRDB) coupled with an adaptive deformable residual convolution mechanism. The RRDB structure effectively mitigates grid artifacts and high-frequency noise by nesting dense connections within a high-level residual framework. (2) To capture complex geomorphic features

adaptive deformable convolutional layers are integrated into the network. This allows the model to adaptively adjust sampling offsets based on local terrain complexity

providing superior flexibility in learning irregular terrain structures. (3) The discriminator is redesigned using a U-Net architecture augmented with an attention block to facilitate multi-scale feature fusion and capture extensive global receptive fields.

Results

The performance is rigorously validated across four distinct terrain datasets. Quantitative analysis indicates that compared to the baseline super-resolution generative adversarial network model

the mean values of the root mean square error and the mean absolute error of the proposed method are reduced by 0.3%-12.6% and 3.2%-14.5%

respectively. Furthermore

terrain derivative accuracy shows marked improvements

with the mean values of the slope error and the aspect error declining by 1.9%-18.3% and 2.7%-7.1%

respectively. Comparative evaluations demonstrate that the proposed SRDCGAN method effectively eliminates visual artifacts and preserves intricate topographic textures.

Conclusions

The experimental findings verify that the proposed SRDCGAN method establishes a robust mapping relationship between high-resolution and low-resolution DEM data

showing strong generalization capabilities across diverse geographical regions.

关键词

Keywords

references

高原，朱娅男，陈传法，等 . 高精度DEM建模的加权径向基函数插值方法［J］. 武汉大学学报（信息科学版）， 2023 ， 48 （ 8 ）： 1373 - 1379 .

GAO Yuan ， ZHU Ya'nan ， CHEN Chuanfa ， et al . A Weighted Radial Basis Function Interpolation Method for High Accuracy DEM Modeling ［J］. Geomatics and Information Science of Wuhan University ， 2023 ， 48 （ 8 ）： 1373 - 1379 .

雷秋佳，刘婧，曹新运 . 利用机载LiDAR数据的开放DEM产品精度评估［J］. 武汉大学学报（信息科学版）， 2025 ， 50 （ 1 ）： 153 - 163 .

LEI Qiujia ， LIU Jing ， CAO Xinyun . Accuracy Evaluation of Open DEM Products Based on Airborne LiDAR Data ［J］. Geomatics and Information Science of Wuhan University ， 2025 ， 50 （ 1 ）： 153 - 163 .

黄文博，张勤，杜源，等 . 融合无人机摄影测量与GNSS的滑坡变形监测方法［J/OL］. 武汉大学学报（信息科学版）， 2025 . https：//doi.org/10.13203/j.whugis20250191 https://doi.org/10.13203/j.whugis20250191 .

HUANG Wenbo ， ZHANG Qin ， DU Yuan ， et al . A Method for Landslide Deformation Monitoring Integrating UAV Photogrammetry and GNSS ［J/OL］. Geomatics and Information Science of Wuhan University ， 2025 . https：//doi.org/10.13203/j.whugis20250191 https://doi.org/10.13203/j.whugis20250191 .

许强，董秀军，赵建军，等 . 基于机载LiDAR的2·8宜宾筠连山体滑坡几何特征分析［J］. 武汉大学学报（信息科学版）， 2025 ， 50 （ 11 ）： 2145 - 2153 .

XU Qiang ， DONG Xiujun ， ZHAO Jianjun ， et al . Geometric Characteristics Analysis of the 2·8 Junlian Landslide， Yibin Based on Airborne LiDAR ［J］. Geomatics and Information Science of Wuhan University ， 2025 ， 50 （ 11 ）： 2145 - 2153 .

LEE S ， WOLBERG G ， SHIN S Y . Scattered Data Interpolation with Multilevel B-Splines ［J］. IEEE Transactions on Visualization and Computer Graphics ， 1997 ， 3 （ 3 ）： 228 - 244 .

刘雄 . 双三次COONS曲面在序列图像超分辨率中的应用［D］. 大连：大连理工大学， 2005 .

LIU Xiong . The Application of Bi-Cubic COONS Surface in Sequence Images Super-Resolution ［D］. Dalian ： Dalian University of Technology ， 2005 .

程勖，张明会 . 基于泛协克里格法的空间插值方法研究［J］. 电脑知识与技术， 2015 ， 11 （ 12 ）： 220 - 222 .

CHENG Xu ， ZHANG Minghui . Research on Spatial Interpolation Methods of Universal Cokriging ［J］. Computer Knowledge and Technology ， 2015 ， 11 （ 12 ）： 220 - 222 .

花卫华，郭丹阳，刘修国，等 . 含复杂倒转的地层层序统一修正与连接方法［J］. 地球科学， 2023 ， 48 （ 4 ）： 1532 - 1542 .

HUA Weihua ， GUO Danyang ， LIU Xiuguo ， et al . Unified Correction and Connection Method of Stratigraphic Sequence with Complex Inversion ［J］. Earth Science ， 2023 ， 48 （ 4 ）： 1532 - 1542 .

花卫华，曾新灵，郭丹阳，等 . 基于构造恢复理论的含复杂断层三维地质建模方法［J］. 地球科学， 2024 ， 49 （ 4 ）： 1411 - 1420 .

HUA Weihua ， ZENG Xinling ， GUO Danyang ， et al . 3D Geological Modeling Method Based on Tectonic Restoration Theory ［J］. Earth Science ， 2024 ， 49 （ 4 ）： 1411 - 1420 .

DONG C ， LOY C C ， HE K M ， et al . Learning a Deep Convolutional Network for Image Super-Resolution ［C］// European Conference on Computer Vision ， Zurich， Switzerland ， 2014 .

陈行，罗斌 . 利用动态上采样滤波深度网络进行多角度遥感影像超分辨率重建［J］. 武汉大学学报（信息科学版）， 2021 ， 46 （ 11 ）： 1716 - 1726 .

CHEN Hang ， LUO Bin . Multi-angle Remote Sensing Images Super-Resolution Reconstruction Using Dynamic Upsampling Filter Deep Network ［J］. Geomatics and Information Science of Wuhan University ， 2021 ， 46 （ 11 ）： 1716 - 1726 .

LEDIG C ， THEIS L ， HUSZÁR F ， et al . Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network ［C］// IEEE Conference on Computer Vision and Pattern Recognition ， Honolulu， USA ， 2017 .

DEMIRAY B Z ， SIT M ， DEMIR I . D-SRGAN： DEM Super-Resolution with Generative Adversarial Networks ［J］. SN Computer Science ， 2021 ， 2 （ 1 ）： 48 .

ZHOU A N ， CHEN Y M ， WILSON J P ， et al . A Multi-terrain Feature-Based Deep Convolutional Neural Network for Constructing Super-Resolution DEMs ［J］. International Journal of Applied Earth Observation and Geoinformation ， 2023 ， 120 ： 103338 .

ZHANG Y F ， YU W H ， ZHU D . Terrain Feature-Aware Deep Learning Network for Digital Elevation Model Superresolution ［J］. ISPRS Journal of Photogrammetry and Remote Sensing ， 2022 ， 189 ： 143 - 162 .

ZHANG Y L ， LI K P ， LI K ， et al . Image Super-Resolution Using Very Deep Residual Channel Attention Networks ［C］// European Conference on Computer Vision ， Munich， Germany ， 2018 .

ZHU D ， CHENG X M ， ZHANG F ， et al . Spatial Interpolation Using Conditional Generative Adversarial Neural Networks ［J］. International Journal of Geographical Information Science ， 2020 ， 34 （ 4 ）： 735 - 758 .

WANG X T ， YU K ， WU S X ， et al . ESRGAN： Enhanced Super-Resolution Generative Adversarial Networks ［C］// European Conference on Computer Vision ， Munich， Germany ， 2018 .

WEI Z H ， HUANG Y D ， CHEN Y A ， et al . A-ESRGAN： Training Real-World Blind Super-Resolution with Attention U-Net Discriminators ［C］// Pacific Rim International Conference on Artificial Intelligence ， Jakarta， Indonesia ， 2023 .

KRISHNAN K S ， KRISHNAN K S . SwiftSRGAN： Rethinking Super-Resolution for Efficient and Real-Time Inference ［C］// International Conference on Intelligent Cybernetics Technology and Applications ， Online ， 2021 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Geometric and Frequency-Domain Constrained Single-Step Diffusion for Seabed Terrain Super-Resolution Reconstruction

LiDAR Point-Cloud Density Optimization for Geohazard Surveys in Densely Vegetated Mountainous Areas

Spatial Super-Resolution Reconstruction of Remote Sensing Precipitation Products Using GAN by Multi-source Fused Precipitation Data Features

BiP-GAN Pedestrian Video Anomaly Event Automatic Detection

Accuracy Evaluation of Open DEM Products Based on Airborne LiDAR Data

Related Author

YANG Fanlin

LIU Yanxiong

BU Xianhai

BI Dekai

PAN Xingyu

WU Fu

LIAO Zeyuan

HE Na

Related Institution

The First Institute of Oceanography, Ministry of Natural Resources

Key Laboratory of Oceanic Surveying and Mapping, Ministry of Natural Resources

College of Geodesy and Geomatics, Shandong University of Science and Technology

State Key Laboratory of Disaster Prevention and Ecology Protection in Open‑Pit Coal Mines, Shandong University of Science and Technology

State Key Laboratory of Geohazrd Prevention and Geoenvironment Protection, Chengdu University of Technology

AI问答

⁰