标题: A Transformer-Based Residual Attention Network Combining SAR and Terrain Features for DEM Super-Resolution Reconstruction
作者: Chen, RX (Chen, Ruoxuan); Chen, YM (Chen, Yumin); Zhang, TF (Zhang, Tengfei); Zeng, F (Zeng, Fei); Li, ZH (Li, Zhanghui)
来源出版物: REMOTE SENSING 卷: 17 期: 21 文献号: 3625 DOI: 10.3390/rs17213625 Published Date: 2025 NOV 1
摘要: Highlights What are the main findings? The incorporation of SAR features and terrain features provides a supplementary data source independent of DEM data, restoring terrain details and enhancing the topographic consistency of super-resolution DEMs. The lightweight Transformer module is combined with the residual feature aggregation structure to enhance global perception capability and reduce redundant model parameters. What are the implications of the main findings? It provides new insights into the research of super-resolution reconstruction of DEMs by integrating multi-source data. It effectively addresses the coexisting challenges of limited global context modeling capability and high computational complexity in deep neural networks.Highlights What are the main findings? The incorporation of SAR features and terrain features provides a supplementary data source independent of DEM data, restoring terrain details and enhancing the topographic consistency of super-resolution DEMs. The lightweight Transformer module is combined with the residual feature aggregation structure to enhance global perception capability and reduce redundant model parameters. What are the implications of the main findings? It provides new insights into the research of super-resolution reconstruction of DEMs by integrating multi-source data. It effectively addresses the coexisting challenges of limited global context modeling capability and high computational complexity in deep neural networks.Abstract Acquiring high-resolution digital elevation models (DEMs) over across extensive regions remains challenging due to high costs and insufficient detail, creating demand for super-resolution (SR) techniques. However, existing DEM SR methods still rely on limited data sources and often neglect essential terrain features. To address the issues, SAR data complements existing sources with its all-weather capability and strong penetration, and a Transformer-based Residual Attention Network combining SAR and Terrain Features (TRAN-ST) is proposed. The network incorporates intensity and coherence as SAR features to restore the details of the high-resolution DEMs, while slope and aspect constraints in the loss function enhance terrain consistency. Additionally, it combines the lightweight Transformer module with the residual feature aggregation module, which enhances the global perception capability while aggregating local residual features, thereby improving the reconstruction accuracy and training efficiency. Experiments were conducted on two DEMs in San Diego, USA, and the results show that compared with methods such as the bicubic, SRCNN, EDSR, RFAN, HNCT methods, the model reduces the mean absolute error (MAE) by 2-30%, the root mean square error (RMSE) by 1-31%, and the MAE of the slope by 2-13%, and it reduces the number of parameters effectively, which proves that TRAN-ST outperforms current typical methods.
作者关键词: digital elevation model; super-resolution reconstruction; SAR features; terrain features
地址: [Chen, Ruoxuan; Chen, Yumin; Zhang, Tengfei; Li, Zhanghui] Wuhan Univ, Sch Resource & Environm Sci, Minist Educ, 129 Luoyu Rd, Wuhan 430079, Peoples R China.
[Zeng, Fei] 32004 Reserve Grp Informat Support Force Peoples L, Wuhan 430079, Peoples R China.
通讯作者地址: Chen, YM (通讯作者)，Wuhan Univ, Sch Resource & Environm Sci, Minist Educ, 129 Luoyu Rd, Wuhan 430079, Peoples R China.
电子邮件地址: [email protected]; [email protected]; [email protected]; [email protected]
影响因子：4.1
