Deep Neural Network Regression for Normalized Digital Surface Model Generation with Sentinel-2 Imagery

Deep Neural Network Regression for Normalized Digital Surface Model Generation with Sentinel-2 Imagery

November 15, 2023

Our EAGLE student Konstantin Müller published together with our chairholder of the Deparment of Global Urbanization and Remote Sensing, Hannes Taubenboeck an article about DL for surface model generation. The article explores methods to extract high-resolution normalized digital surface models (nDSMs) from low-resolution Sentinel-2 data, enabling the creation of large-scale models. Leveraging the open access and global coverage of Sentinel 2, the study employs deep learning models, based on the U-Net architecture, with tailored multiscale encoders and conformed self-attention to achieve a mean height error of approximately 2m.

from the abstract: In recent history, normalized digital surface models (nDSMs) have been constantly gaining importance as a means to solve large-scale geographic problems. High-resolution surface models are precious, as they can provide detailed information for a specific area. However, measurements with a high resolution are time consuming and costly. Only a few approaches exist to create high-resolution nDSMs for extensive areas. This article explores approaches to extract high-resolution nDSMs from lowresolution Sentinel-2 data, allowing us to derive large-scale models. We thereby utilize the advantages of Sentinel 2 being open access, having global coverage, and providing steady updates through a high repetition rate. Several deep learning models are trained to overcome the gap in producing high-resolution surface maps from low-resolution input data. With U-Net as a base architecture, we extend the capabilities of our model by integrating tailored multiscale encoders with differently sized kernels in the convolution as well as conformed self-attention inside the skip connection gates. Using pixelwise regression, our U-Net base models can achieve a mean height error of approximately 2 m. Moreover, through our enhancements to the model architecture, we reduce the model error by more than 7%.

Link: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10189905

you may also like:

Presentations at the EARSeL conference in Manchester

Presentations at the EARSeL conference in Manchester

Presentations at the EARSeL conference in Manchester   Dr. Marta Sapena and Dr. John Friesen represented the Earth Observation Center (EOC) of the German Aerospace Center (DLR) and our Earth Observation Research Cluster (EORC) this week at the EARSeL conference...

Our PhD candidate Ines Standfuss teaches at AniMove

Our PhD candidate Ines Standfuss teaches at AniMove

Our PhD candidate Ines Standfuss is teaching remote sensing for animal movement analysis this year at MPI at Lake Constanze. The AniMove science school has been founded more than ten years ago together with MPI and other organisations such as Smithsonian joined in the...

television and radio coverage about urban measurements

television and radio coverage about urban measurements

Our urban research got covered by TV and radio where we had the chance to explain the relevance of urban monitoring via remote sensing methods as well as in-situ devices (in cooperation with Prof. Marco Schmidt) especially for adaptation and mitigation potential of...

“Super Test Site Würzburg” – from the idea to realization

“Super Test Site Würzburg” – from the idea to realization

The "Super Test Site Würzburg" originated as an idea at the "Geolingual Studies Workshop", which our Geolingual Studies team (Prof. Biewer, Prof. Taubenböck) organized last year - we had reported on it: https://remote-sensing.org/geolingual-studies-workshop-day-1-2/ ;...

Two new Msc graduates Caroline and Helena

Two new Msc graduates Caroline and Helena

We are very happy to congratulate Caroline Goehner and Helena Wehner for successfully defending their thesis on alpine research and ibis environment analysis.  Great to see their journey through the eagle program, their technical advances, gaining hands-on...

Science and teaching at UFS

Science and teaching at UFS

The research and teaching activities of our University at the research station Schneefernerhaus on Zugspitze are increasing steadily and becoming more diverse. Ongoing and planned courses and science projects range from topics in informatics, astronomy to medicine and...

Privacy Policy

Lehrstuhl für Fernerkundung & Lehrstuhl für Urbane Fernerkundung

Erdbeobachtung an der Universität Würzburg