Deep gradient prior network for DEM super-resolution: Transfer learning from image to DEM

Xu, Zhenlin; Chen, Zixuan; Yi, Weiwei; Gui, Qiuling; Hou, Wenguang; Ding, Mingyue

doi:10.1016/j.isprsjprs.2019.02.008

Cited by 50 publications

(45 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The variability in the quality of openly accessible DEMs suggests that these openly accessible DEMs should be utilized cautiously as per the application requirements. New techniques also demand correct selection of input DEMs, while executing DEM fusion or generation of superresolution DEMs providing an opportunity to improve DEMs [10,11,17].…”

Section: Discussionmentioning

confidence: 99%

“…that create ideal conditions for remote sensing [7]. TanDEM-X has been used for applications like extraction of digital building height models [8], archaeological sites [9], DEM fusion using ANN techniques [10] and DEM super-resolution [11]. AW3D30 was found to be the most promising while investigating the performances of seven public freely-accessed DEM datasets (ASTER GDEM V2, SRTM-3 V4.1 DEM, SRTM-1 DEM, AW3D30 DEM, VFP-DEM, MERIT DEM, Seamless SRTM-1 DEM) over the HMA region (Hengduan Mountains and Himalayas) by referring to high-accuracy elevation data from ICESat altimetry [12].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Assessment of Vertical Accuracy for TanDEM-X 90 m DEMs in Plain, Moderate, and Rugged Terrain

Bhardwaj¹

2019

The 2nd International Electronic Conference on Geosciences

View full text Add to dashboard Cite

Synthetic Aperture Radar (SAR) interferometry technique generates digital elevation models (DEMs) and is used by various agencies widely. The recently released TanDEM-X DEM by DLR at 90 m spatial resolution is available for free download to users. This paper examines the accuracy of TanDEM-X DEM at different experimental sites with different topographic characteristics. Three sites were chosen, namely Kendrapara (Odisha), Jaipur (Rajasthan), and Dehradun (Uttarakhand) with plain, moderate, and highly undulating terrain conditions. The root mean square error (RMSE) were calculated using ground control points (GCPs) collected by differential GPS method for experimental sites at Dehradun, Jaipur, and Kendrapara. The accuracy of TanDEM-X 90 m datasets is compared with other openly accessible optically-derived DEMs (ASTER GDEM V2, CartoDEM V3 R1, AW3D30) and InSAR-derived DEMs (SRTM, ALOS PALSAR RTC HR). The RMSEs reveal that at Jaipur site with moderate terrain with urban and agriculture as major land use land cover (LULC) classes, the results of TanDEM-X 90 m DEM have higher accuracy than ALOS PALSAR RTC HR DEM. However, it is observed that in a predominantly plain region with agriculture practice (Kendrapara site, Odisha) and rugged region (Dehradun site, Uttarakhand) with mixed land use land cover (LULC) (e.g., forest, urban, streams, and agriculture) the results of ALOS PALSAR RTC HR data have higher accuracy than TanDEM-X 90 m DEM. Further, the study indicates that for a relatively plain site at Kendrapara (Orissa), CartoDEM V3 R1 DEM has the best performance with an RMSE of 1.96 m, which is the lowest among all DEMs utilized in the study.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Assessment of Vertical Accuracy for TanDEM-X 90 m DEMs in Plain, Moderate, and Rugged Terrain

Bhardwaj¹

2019

The 2nd International Electronic Conference on Geosciences

View full text Add to dashboard Cite

show abstract

“…Nevertheless, directly utilizing image ESRGAN to the case of DEM is uncomplicated in theory but hard in practice. Most natural images are of 8-bit sizes, whose grey range is usually between 0 and 255, while the range of most DEM images exceeds 255 greatly (Xu et al, 2019). Thus, if training an ESRGAN for DEM SR directly, we need a considerable amount of DEM samples consist of vastly diverse height values.…”

Section: Methodsmentioning

confidence: 99%

Esrgan-Based Dem Super-Resolution for Enhanced Slope Deformation Monitoring in Lantau Island of Hong Kong

Ma²

2020

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Monitoring, evaluating and understanding the slopes by Interferometric Synthetic Aperture Rader (InSAR) technology are critical for both human economy and natural environment. However, the resolution limitation of existing digital elevation model (DEM) in the slope areas causes the DEM phase residues and atmospheric effects promoted, which will influence the interpret accuracy of InSAR results. In this study, we propose a novel two-step ESRGAN-based DEM SR method to effectively recover high-resolution DEM from the original version. Firstly, we pretrain an ESRGAN with a large number of natural images. Based on it, we transfer the learnt knowledge into the DEM problem and fine-tune the DEM SR network. The recovered DEMs are utilized as the reference data to improve slope deformation monitoring and enhance the accuracy of InSAR estimation, especially in the mountainous areas with cloudy and rainy weather. Experiments indicate that the proposed method can achieve better results than the traditional methods and works in phase simulation, which is one of the key step of InSAR deformation monitoring.

show abstract

“…Nevertheless, the required extra-high-accuracy topographic information for the implementation of this type of method is still hard to acquire, especially for a large extent. The learning-based approaches generate high-resolution DEMs by establishing the correlation between low-and high-resolution DEMs through a training process [29][30][31][32][33]. Learning-based models can be trained to learn from multi-dimensional information, which may potentially produce high-resolution DEMs of better quality.…”

Section: Introductionmentioning

confidence: 99%

A Multi-Scale Mapping Approach Based on a Deep Learning CNN Model for Reconstructing High-Resolution Urban DEMs

Jiang

Xia

et al. 2020

Water

View full text Add to dashboard Cite

The scarcity of high-resolution urban digital elevation model (DEM) datasets, particularly in certain developing countries, has posed a challenge for many water-related applications such as flood risk management. A solution to address this is to develop effective approaches to reconstruct high-resolution DEMs from their low-resolution equivalents that are more widely available. However, the current high-resolution DEM reconstruction approaches mainly focus on natural topography. Few attempts have been made for urban topography, which is typically an integration of complex artificial and natural features. This study proposed a novel multi-scale mapping approach based on convolutional neural network (CNN) to deal with the complex features of urban topography and to reconstruct high-resolution urban DEMs. The proposed multi-scale CNN model was firstly trained using urban DEMs that contained topographic features at different resolutions, and then used to reconstruct the urban DEM at a specified (high) resolution from a low-resolution equivalent. A two-level accuracy assessment approach was also designed to evaluate the performance of the proposed urban DEM reconstruction method, in terms of numerical accuracy and morphological accuracy. The proposed DEM reconstruction approach was applied to a 121 km2 urbanized area in London, United Kingdom. Compared with other commonly used methods, the current CNN-based approach produced superior results, providing a cost-effective innovative method to acquire high-resolution DEMs in other data-scarce regions.

show abstract

Deep gradient prior network for DEM super-resolution: Transfer learning from image to DEM

Cited by 50 publications

References 22 publications

Assessment of Vertical Accuracy for TanDEM-X 90 m DEMs in Plain, Moderate, and Rugged Terrain

Assessment of Vertical Accuracy for TanDEM-X 90 m DEMs in Plain, Moderate, and Rugged Terrain

Esrgan-Based Dem Super-Resolution for Enhanced Slope Deformation Monitoring in Lantau Island of Hong Kong

A Multi-Scale Mapping Approach Based on a Deep Learning CNN Model for Reconstructing High-Resolution Urban DEMs

Contact Info

Product

Resources

About