A global coastal DEM super-resolution reconstruction method integrating frequency-domain features and topographic priors

doi:10.11947/j.AGCS.2025.20250126

Abstract

Abstract:

To overcome the limitations of existing digital elevation model (DEM) super-resolution reconstruction methods in model generalization and terrain feature awareness, a terrain-factor-guided approach is proposed. First, a paired high- and low-resolution DEM dataset covering global coastal regions is constructed based on the open-source DEM data from GEBCO and ETOPO-1. Then, an end-to-end DEM super-resolution model is designed with a high-low frequency feature decoupling module to separate and fuse terrain details and global features. A terrain guidance module is introduced, incorporating prior terrain knowledge, such as surface segmentation depth, to enhance the perception of terrain structure. Additionally, dynamic weight adjustment based on homoscedastic uncertainty is employed to adaptively assign loss weights according to the importance of different terrain features, thereby improving reconstruction accuracy and stability. Comparative experiments with seven state-of-the-art methods demonstrate that the proposed approach achieves superior performance in RMSE_elevation, MAE_elevation, and RMSE_slope metrics, accurately reconstructing complex features such as fragmented ridges and low seamounts. These results validate the effectiveness and robustness of the method in terrain feature reconstruction and precision enhancement.

Key words: digital elevation model, super-resolution reconstruction, terrain factors, prior-guided, feature fusion

CLC Number:

P208

Wenjun HUANG, Qun SUN, Qing XU, Long FAN, Anzhu YU, Fubing ZHANG. A global coastal DEM super-resolution reconstruction method integrating frequency-domain features and topographic priors[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(8): 1518-1531.

Figures/Tables 12

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Tab. 1

Tab. 2

Tab. 3

Tab. 4

References 34

[1]	汤国安, 李发源, 刘学军. 数字高程模型教程[M]. 3版. 北京: 科学出版社, 2016.
	TANG Guoan, LI Fayuan, LIU Xuejun. Digital elevation model course[M]. 3rd ed. Beijing: Science Press, 2016.
[2]	王光霞, 朱长青, 史文中, 等. 数字高程模型地形描述精度的研究[J]. 测绘学报, 2004, 33(2): 168-173.
	WANG Guangxia, ZHU Changqing, SHI Wenzhong, et al. The further study on the accuracy of DEM terrain representation[J]. Acta Geodaetica et Cartographica Sinica, 2004, 33(2): 168-173.
[3]	汤国安, 那嘉明, 程维明. 我国区域地貌数字地形分析研究进展[J]. 测绘学报, 2017, 46(10): 1570-1591. DOI: . doi: 10.11947/j.AGCS.2017.20170388
	TANG Guoan, NA Jiaming, CHENG Weiming. Progress of digital terrain analysis on regional geomorphology in China[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(10): 1570-1591. DOI: . doi: 10.11947/j.AGCS.2017.20170388
[4]	YANG Jiaqi, XU Jun, ZHU Yunqiang, et al. GeomorPM: a geomorphic pretrained model integrating convolution and Transformer architectures based on DEM data[J]. International Journal of Geographical Information Science, 2025, 39(2): 422-451.
[5]	AVAND M, KURIQI A, KHAZAEI M, et al. DEM resolution effects on machine learning performance for flood probability mapping[J]. Journal of Hydro-Environment Research, 2022, 40: 1-16.
[6]	赵建虎, 欧阳永忠, 王爱学. 海底地形测量技术现状及发展趋势[J]. 测绘学报, 2017, 46(10): 1786-1794. DOI: . doi: 10.11947/j.AGCS.2017.20170276
	ZHAO Jianhu, OUYANG Yongzhong, WANG Aixue. Status and development tendency for seafloor terrain measurement technology[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(10): 1786-1794. DOI: . doi: 10.11947/j.AGCS.2017.20170276
[7]	李鹏. 海岸带地理环境雷达遥感监测关键问题研究[J]. 测绘学报, 2021, 50(4): 565. DOI: . doi: 10.11947/j.AGCS.2021.20200165
	LI Peng. Key issues on coastal geographical environment monitoring with radar remote sensing[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(4): 565. DOI: . doi: 10.11947/j.AGCS.2021.20200165
[8]	KARIMINEJAD N, HOSSEINALIZADEH M, POURGHASEMI H R, et al. Optimizing collapsed pipes mapping: effects of DEM spatial resolution[J]. Catena, 2020, 187: 104344.
[9]	王耀革, 王鑫, 朱长青. 基于双线性内插规则格网DEM地形误差模型[J]. 测绘科学技术学报, 2007, 24(6): 419-421.
	WANG Yaoge, WANG Xin, ZHU Changqing. A terrain error model of grid DEM based on bilinear polynomial[J]. Journal of Zhengzhou Institute of Surveying and Mapping, 2007, 24(6): 419-421.
[10]	KEYS R G. Cubic convolution interpolation for digital image processing[J]. IEEE Transactions on Acoustics Speech and Signal Processing, 1981, 29: 1153-1160.
[11]	郭伟伟, 章品正. 基于迭代反投影的超分辨率图像重建[J]. 计算机科学与探索, 2009, 3(3): 321-329.
	GUO Weiwei, ZHANG Pinzheng. Super-resolution image reconstruction with iterative back projection algorithm[J]. Journal of Frontiers of Computer Science & Technology, 2009, 3(3): 321-329.
[12]	PROTTER M, ELAD M, TAKEDA H, et al. Generalizing the nonlocal-means to super-resolution reconstruction[J]. IEEE Transactions on Image Processing, 2009, 18(1): 36-51.
[13]	DONG Chao, LOY C C, HE Kaiming, et al. Image super-resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2): 295-307.
[14]	LIM B, SON S, KIM H, et al. Enhanced deep residual networks for single image super-resolution[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops. Honolulu: IEEE, 2017: 136-144.
[15]	LEDIG C, THEIS L, HUSZÁR F, et al. Photo-realistic single image super-resolution using a generative adversarial network[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 105-114.
[16]	WANG Xintao, YU Ke, WU Shixiang, et al. ESRGAN: enhanced super-resolution generative adversarial networks[C]//Proceedings of 2018 European Conference on Computer Vision. Cham: Springer, 2019: 63-79.
[17]	ZHANG Yifan, YU Wenhao, ZHU Di. Terrain feature-aware deep learning network for digital elevation model superresolution[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2022, 189: 143-162.
[18]	DAI Jifeng, QI Haozhi, XIONG Yuwen, et al. Deformable convolutional networks[C]//Proceedings of 2017 IEEE International Conference on Computer Vision. Venice: IEEE, 2017: 764-773.
[19]	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/OL]. 武汉大学学报(信息科学版). [2025-02-20]. https://doi.org/10.13203/j.whugis20240186.
	HOU Xujuan, DENG Xiaotong, HUA Weihua, et al. DEM super-resolution reconstruction method based on adaptive generative adversarial network[J/OL]. Geomatics and Information Science of Wuhan University. [2025-02-20]. https://doi.org/10.13203/j.whugis20240186.
[20]	陈凯, 雷少华, 代文, 等. 基于开源数据和条件生成对抗网络的地形重建方法[J]. 地球信息科学学报, 2023, 25(2): 252-264.
	CHEN Kai, LEI Shaohua, DAI Wen, et al. Terrain rebuilding method based on open source data and conditional generative adversarial networks[J]. Journal of Geo-information Science, 2023, 25(2): 252-264.
[21]	CAI Wuxu, LIU Yanxiong, CHEN Yilan, et al. A seabed terrain feature extraction transformer for the super-resolution of the digital bathymetric model[J]. Remote Sensing, 2023, 15(20): 4906.
[22]	WANG Yi, JIN Shichao, YANG Zekun, et al. TTSR: a Transformer-based topography neural network for digital elevation model super-resolution[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 3360489.
[23]	TOZER B, SANDWELL D T, SMITH W H F, et al. Global bathymetry and topography at 15 arcsec: SRTM15+[J]. Earth and Space Science, 2019, 6(10): 1847-1864.
[24]	AMANTE C, EAKINS B W. ETOPO1 arc-minute global relief model: procedures, data sources and analysis[R]. Boulder: National Geophysical Data Center, 2009.
[25]	HENDRYCKS D, GIMPEL K. Gaussian error linear units (GELUs)[EB/OL]. [2025-02-20]. https://arxiv.org/abs/1606.08415.
[26]	IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]//Proceedings of 2015 International Conference on Machine Learning. Lille: PMLR, 2015: 448-456.
[27]	ULYANOV D, VEDALDI A, LEMPITSKY V. Improved texture networks: maximizing quality and diversity in feed-forward stylization and texture synthesis[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 4105-4113.
[28]	SALIMANS T, KINGMA D P. Weight normalization: a simple reparameterization to accelerate training of deep neural networks[C]//Proceedings of the 30th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2016: 901-909.
[29]	CIPOLLA R, GAL Y, KENDALL A. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 7482-7491.
[30]	LIANG Jingyun, CAO Jiezhang, SUN Guolei, et al. SwinIR: image restoration using swin transformer[C]//Proceedings of 2021 IEEE/CVF International Conference on Computer Vision Workshops. Montreal: IEEE, 2021: 1833-1844.
[31]	ZHOU Yupeng, LI Zhen, GUO Chunle, et al. SRFormer: permuted self-attention for single image super-resolution[C]//Proceedings of 2023 IEEE/CVF International Conference on Computer Vision. Paris: IEEE, 2023: 12734-12745.
[32]	LIU Hongying, LI Zekun, SHANG Fanhua, et al. Arbitrary-scale super-resolutionvia deep learning: a comprehensive survey[J]. Information Fusion, 2024, 102: 102015.
[33]	孙群, 温伯威, 陈欣. 多源地理空间数据一致性处理研究进展[J]. 测绘学报, 2022, 51(7): 1561-1574. DOI: . doi: 10.11947/j.AGCS.2022.20220151
	SUN Qun, WEN Bowei, CHEN Xin. Research on consistency processing of multi-source geospatial data[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(7): 1561-1574. DOI: . doi: 10.11947/j.AGCS.2022.20220151
[34]	孙群, 任福, 龙毅, 等. 中国地图制图学与地理信息工程进展报告(2019—2023)[J]. 测绘学报, 2024, 53(3): 399-412. DOI: . doi: 10.11947/j.AGCS.2024.20230562
	SUN Qun, REN Fu, LONG Yi, et al. The progress and trend of cartography and geographic information engineering in China (2019—2023)[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(3): 399-412. DOI: . doi: 10.11947/j.AGCS.2024.20230562

方法	RMSE_elevation/m	MAE_elevation/m	RMSE_slope/(°)
双三次插值	29.084	22.556	5.234
ESRGAN	27.385	20.252	4.788
EDSR	27.259	20.654	4.754
SwinIR	27.220	20.418	4.723
SRFormer	26.286	19.455	4.678
TfaSR	25.991	19.239	4.347
StfeT	26.803	19.642	4.691
本文方法	25.543	19.132	4.289

方法	RMSE_elevation/m	MAE_elevation/m	RMSE_slope/(°)
HLF-FDM去掉高频分支	26.148	19.401	4.526
HLF-FDM去掉低频分支	25.821	19.334	4.315
去掉TGM	26.765	19.682	4.681
本文方法	25.543	19.132	4.289

迭代次数	RMSE_elevation/m	MAE_elevation/m	RMSE_slope/(°)	FLOPs	参数量
5	26.118	19.451	4.598	2.556×10⁶	98.438
6(本文方法)	25.543	19.132	4.289	3.000×10⁶	112.988
7	25.776	19.284	4.230	3.444×10⁶	127.538

损失函数			RMSE_elevation/m	MAE_elevation/m	RMSE_slop_e/(°)
			RMSE_elevation/m	MAE_elevation/m	RMSE_slop_e/(°)
√			26.261	19.423	4.639
√	√		25.782	19.308	4.517
√	√	√	25.543	19.132	4.289