We developed a United U-Net (UU-Net) to fuse optical and SAR data for road extraction, which was trained and evaluated on a large-scale multisource road extraction dataset. The UU-Net achieved better accuracy than traditional deep convolutional networks with optical or SAR data alone, which ...
from optical images. As a large open SAR-optical dataset with multiple scenes of a high resolution, we believe QXS-SAROPT will be of potential value for further research in SAR-optical data fusion technology based on deep learning. The QXS-SAROPT Dataset for Deep Learning in SAR-Optical ...
This repo contains the model code, written in Python/Keras, as well as links to pre-trained checkpoints and the SEN12MS-CR dataset. deep-learning satellite sentinel satellite-imagery satellite-data sar optical sentinel-2 residual-neural-network sentinel-1 cloud-removal Updated May 22, 2024 ...
Nodes with standard deviation (STDV) larger than 10 m are removed from the dataset as they are suspected to be not related to the tectonic signal, or poorly reliable due to high signal to noise ratio. At the end of the post-processing we get 10313 surface nodes for S1 correlograms (...
THE SEN1-2 DATASET FOR DEEP LEARNING IN SAR-OPTICAL DATA FUSIONM. Schmitt 1 , L. H. Hughes 1 , X. X. Zhu 1,21 Signal Processing in Earth Observation, Technical University of Munich (TUM), Munich, Germany - (m.schmitt,lloyd.hughes)@tum.de2 Remote Sensing Technology Institute (IMF)...
• A modified image translation GAN architecture with multiscale cascaded residual connections is proposed for raw image translation between two very different sensing modalities, SAR and optical sensors. • Experiment results on large volume of dataset demonstrate good visual quality and variety that...
Table 1 Detailed description of dataset. Full size table Figure 1 (a) Optical image; (b) SAR image; Matches found in pair 1 using (c) PSO-SIFT, (d) OS-SIFT, and (e) the proposed method. The reference image is shown on the left and the sensed image on the right. Full size imag...
3.1. SAR and optical correspondence To train the correspondence network we require a large dataset of salient candidate search and template patches with known points of correspondence. Due to the complexity of creating such a dataset, and the intractability of manually annotating correspondence across ...
3. Department of Biomedical Engineering, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong SAR, People’s Republic of China More Information Abstract Abstract The growing demand for electronic devices, smart devices, and the Internet of Things constitutes the primary driving for...
\\{SAR\\}LandsatAVNIR-2TextureThe classification of tropical fragmented landscapes and moist forested areas is a challenge due to the presence of a continuum of vegetation successional stages, persistent cloud cover and the presence of small patches of different land cover types. To classify one ...