Image Restoration from Limited Data
Abstract
Ground-based imagery of satellites is a cornerstone of SSA. The resolution of this imagery is fundamentally limited by turbulence in the atmosphere. Full resolution can be restored by using advanced multi-frame blind deconvolution (MFBD) algorithms which, applied to sequences of short-exposure images, estimate the object scene and point spread functions (PSFs) that characterize the turbulence. Because there are always more variables to estimate than measurements, MFBD is an ill-posed problem. Furthermore, in the regime of limited data, for example a satellite with a rapidly changing pose, the problem is also ill-conditioned because of the lack of diversity in the PSFs. These challenges typically lead to poor quality restorations. The Daylight Object Restoration Algorithm (DORA) overcomes this problem, by using additional simultaneous measurements from a wave-front sensor, along with a frozen flow model of the atmosphere, to achieve high-resolution estimates of space objects from limited data sets. The improvement in image resolution achieved by DORA when compared to current state of the art MFBD algorithms is demonstrated using real data.
- Publication:
-
Advanced Maui Optical and Space Surveillance (AMOS) Technologies Conference
- Pub Date:
- 2017
- Bibcode:
- 2017amos.confE..86H
- Keywords:
-
- Imaging;
- Image Restoration