Informatics
Earth System Sciences
Mar 1, 2018

Super-resolution image reconstruction for PROBA-V satellite

Proba-V satellite
Proba-V satellite

Due to theoretical, practical, and cost constraints, every satellite imaging system produces images with limited spatial resolution. Nevertheless, it is possible to enhance the resolution of the images acquired by means of image post-processing algorithms. Super-resolution image reconstruction has been a very active research area in the past couple of decades [1]. One approach often used is the so-called multi-frame super-resolution, where a set of low-resolution images of the same scene are combined in order to create a single higher-resolution image [2]. Many new satellites are taking advantage of such super-resolution algorithms, including the SkySat-1, the first microsatellite-class commercial earth observation system to produce sub-meter resolution imagery [3].

The ACT is exploring the possibility to apply super-resolution image reconstruction to imagery of an already operational satellite PROBA-V, a vegetation observation satellite that covers most of the Earth's surface every day. Assuming that Earth's vegetation does not change considerably on a daily basis, it might be possible to fuse multiple consecutive images of the same area, collected over the course of several days or even weeks, in order to produce imagery of vegetation with enhanced spatial detail.

References

  1. S. C. Park, M. K. Park, M. G. Kang, "Super-Resolution Image Reconstruction: A Technical Overview", IEEE Signal Processing Magazine, Vol. 20(3), 2003.
  2. S. Farsiu, D. Robinson, M. Elad, P. Milanfar, "Fast and Robust Multiframe Super Resolution", IEEE Transactions on Image Processing, Vol. 13(10), 2004.
  3. K. Murthy, M. Shearn, B. D. Smiley, A. H. Chau, J. Levine, and M. D. Robinson, "SkySat-1: Very High-Resolution Imagery from a Small Satellite", Proc. of SPIE, Vol. 9241, 2014.

Outcome

Artificial Intelligence Peer reviewed article
Super-resolution of PROBA-V images using convolutional neural networks
Märtens, Marcus and Izzo, Dario and Krzic, Andrej and Cox, Daniel
Astrodynamics
(2019)
Download
BibTex
Hamburger icon
Menu
Advanced Concepts Team