About the research

Remote sensing is the science of acquiring information about objects from a distance. It has proven to be crucial for mapping and monitoring of the environment. Optical remote sensing is based on using sensors that are sensitive to wavelengths ranging from 400 to 2500 nm, i.e., from the visible to the near- and the mid-infrared portion of the electromagnetic spectrum. Multispectral systems are based on multichannel detectors with few spectral bands. However, due to technological constraints, data transmission limitations, and signal to noise ratio (SNR) issues many multispectral systems have a spatial resolution that varies between bands. Example of such system is the recently deployed Sentinel-2 (S2) constellation. The S2 data is publicly available, and there is a huge interest in using this data within the research community.

For scientific use of this data it is very beneficial to sharpen the low-resolution bands of S2. In this project, we will develop both model-based and deep learning based methods for sharpening S2 data. We will furthermore develop tuning parameter selection methods and extend the methods to the hypersharpening and pansharpening problems. Furthermore, we will develop an end-to-end combined sharpening and hyperspectral unmixing system. The techniques will be implemented in Python and Matlab and made available to the research community.

See further information