image registration

Data-driven registration of satellite data

The immense availability of earth observation data has enabled the remote sensing community to make great progress on a variety of applications. More often than not, these applications require that the employed images are on the same coordinate system, making the registration process a necessary prerequisite.

A schematic diagram of the image matching and registration presented in Deep Learning for the Earth Sciences [Vakalopoulou et al. 2021]
  • In Papadomanolaki et al. 2021 we present an end-to-end multistep approach for learning an accurate mapping between unregistered image pairs of satellite data.
Results of the method presented on the MDPI 2021 [Papadomanolaki et al. 2021]

References

2021

  1. Unsupervised multistep deformable registration of remote sensing imagery based on deep learning
    Maria Papadomanolaki, Stergios Christodoulidis, Konstantinos Karantzalos, and 1 more author
    Remote Sensing, 2021
  2. Deep learning for image matching and Co-registration
    Maria Vakalopoulou, Stergios Christodoulidis, Mihir Sahasrabudhe, and 1 more author
    Deep Learning for the Earth Sciences: A Comprehensive Approach to Remote Sensing, Climate Science, and Geosciences, 2021