A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMs
Scale conversion between DEMs is an important issue in geomorphometry. There are many mature studies on the generation of low-resolution(LR) DEMs from high-resolution(HR) DEMs. However, as an important and convenient means of obtaining HR DEMs, traditional super resolution (SR) methods have shown in...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-06-01
|
| Series: | International Journal of Applied Earth Observations and Geoinformation |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1569843223001607 |
| _version_ | 1797816076465078272 |
|---|---|
| author | Annan Zhou Yumin Chen John P. Wilson Guodong Chen Wankun Min Rui Xu |
| author_facet | Annan Zhou Yumin Chen John P. Wilson Guodong Chen Wankun Min Rui Xu |
| author_sort | Annan Zhou |
| collection | DOAJ |
| description | Scale conversion between DEMs is an important issue in geomorphometry. There are many mature studies on the generation of low-resolution(LR) DEMs from high-resolution(HR) DEMs. However, as an important and convenient means of obtaining HR DEMs, traditional super resolution (SR) methods have shown insufficient consideration of the terrain features embedded in DEMs. Therefore, this article investigates the combination of terrain features and the use of convolutional neural networks (CNN) to reconstruct HR DEMs, and proposes a multi-terrain feature-based deep CNN for super-resolution(SR) DEMs (MTF-SR). In the experiments, from the perspective of vector and raster terrain features, we fuse raster terrain features in the input and loss functions, and fuse vector terrain features in the optimization of the output of the model. The results show that the MTF-SR model has a 30–50 % reduction in mean absolute error (MAE) compared with interpolation methods, has the lowest slope and aspect error and has a 10 to 30 % improvement in streamline matching rate (SMR). These results point to the advantages of the method in overall elevation accuracy and the preservation of terrain features. |
| first_indexed | 2024-03-13T08:32:20Z |
| format | Article |
| id | doaj.art-cc5e7851e3484d5f8685e508c5f26ae5 |
| institution | Directory Open Access Journal |
| issn | 1569-8432 |
| language | English |
| last_indexed | 2024-03-13T08:32:20Z |
| publishDate | 2023-06-01 |
| publisher | Elsevier |
| record_format | Article |
| series | International Journal of Applied Earth Observations and Geoinformation |
| spelling | doaj.art-cc5e7851e3484d5f8685e508c5f26ae52023-05-31T04:43:56ZengElsevierInternational Journal of Applied Earth Observations and Geoinformation1569-84322023-06-01120103338A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMsAnnan Zhou0Yumin Chen1John P. Wilson2Guodong Chen3Wankun Min4Rui Xu5School of Resource and Environment Science, Wuhan University, Wuhan, Hubei 430079, ChinaSchool of Resource and Environment Science, Wuhan University, Wuhan, Hubei 430079, China; Key Laboratory of Geographic Information System, Ministry of Education, Wuhan University, 129 Luoyu Road, Wuhan 430079, China; Key Laboratory of Digital Cartography and Land Information Application, Ministry of Natural Resources of People’s Republic of China, Wuhan University, 129 Luoyu Road, Wuhan 430079, China; Corresponding author at: 129 Luoyu Road, Wuhan, Hubei 430079, China.Spatial Sciences Institute, University of Southern California, Los Angeles, CA 90089-0374, USASchool of Resource and Environment Science, Wuhan University, Wuhan, Hubei 430079, ChinaSchool of Resource and Environment Science, Wuhan University, Wuhan, Hubei 430079, ChinaSchool of Resource and Environment Science, Wuhan University, Wuhan, Hubei 430079, ChinaScale conversion between DEMs is an important issue in geomorphometry. There are many mature studies on the generation of low-resolution(LR) DEMs from high-resolution(HR) DEMs. However, as an important and convenient means of obtaining HR DEMs, traditional super resolution (SR) methods have shown insufficient consideration of the terrain features embedded in DEMs. Therefore, this article investigates the combination of terrain features and the use of convolutional neural networks (CNN) to reconstruct HR DEMs, and proposes a multi-terrain feature-based deep CNN for super-resolution(SR) DEMs (MTF-SR). In the experiments, from the perspective of vector and raster terrain features, we fuse raster terrain features in the input and loss functions, and fuse vector terrain features in the optimization of the output of the model. The results show that the MTF-SR model has a 30–50 % reduction in mean absolute error (MAE) compared with interpolation methods, has the lowest slope and aspect error and has a 10 to 30 % improvement in streamline matching rate (SMR). These results point to the advantages of the method in overall elevation accuracy and the preservation of terrain features.http://www.sciencedirect.com/science/article/pii/S1569843223001607Terrain featuresConvolutional neural networksDigital elevation modelsSuper-resolution |
| spellingShingle | Annan Zhou Yumin Chen John P. Wilson Guodong Chen Wankun Min Rui Xu A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMs International Journal of Applied Earth Observations and Geoinformation Terrain features Convolutional neural networks Digital elevation models Super-resolution |
| title | A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMs |
| title_full | A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMs |
| title_fullStr | A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMs |
| title_full_unstemmed | A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMs |
| title_short | A multi-terrain feature-based deep convolutional neural network for constructing super-resolution DEMs |
| title_sort | multi terrain feature based deep convolutional neural network for constructing super resolution dems |
| topic | Terrain features Convolutional neural networks Digital elevation models Super-resolution |
| url | http://www.sciencedirect.com/science/article/pii/S1569843223001607 |
| work_keys_str_mv | AT annanzhou amultiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT yuminchen amultiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT johnpwilson amultiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT guodongchen amultiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT wankunmin amultiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT ruixu amultiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT annanzhou multiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT yuminchen multiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT johnpwilson multiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT guodongchen multiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT wankunmin multiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems AT ruixu multiterrainfeaturebaseddeepconvolutionalneuralnetworkforconstructingsuperresolutiondems |