Restoration of Multitemporal Short-Exposure Astronomical Images

Abstract

A multitemporal fast adaptive recursive restoration method based on the underlying spatial probabilistic image model is presented. The method assumes linear degradation model with the unknown possibly non-homogeneous point-spread function and additive noise. Pixels in the vicinity of image steep discontinuities are left unrestored to minimize restoration blurring effect. The method is applied for astronomical sunspot image restoration, where for every ideal undegraded unobservable image several degraded observed images are available.

Modern solar facilities - advanced solar science

“…The lighter shades represent higher predictor probabilities while dark areas were not changed during the reconstruction. Visual comparison of the reconstructed images with the input degraded images as well as the criterion values (Haindl &Šimberová 2005) demonstrate clearly the deblurring effect of the presented algorithm and restoration improvement. The proposed method was superior over the classical methods (e.g.…”

“…All images were corrected by the MTF of the telescope, but no aberration and seeing corrections have been applied. As an objective evaluation measure of the restoration performance an integral of a sum of image partial derivatives (Subbarao et al 1993; see results in Haindl &Šimberová 2005), as well as visual evaluation has been used. The solar images in Fig.…”

“…We assume knowledge of all pixels from the reconstructed scene. For the treatment of the more difficult problem when some data are missing see Haindl & Simberová (1996), Haindl &Šimberová (1998. The image degradation is supposed to be approximated by the linear discrete spatial domain degradation model…”

“…Single matrix elements in the expectation E{X} are approximated (Haindl &Šimberová 2005) by the conditional expectation…”

“…The model adaptivity is introduced using the exponential forgetting factor technique in parameter learning part of the algorithm. The conditional mean value can be derived under few acceptable conditions (Haindl &Šimberová 2005) in the following form:…”