Astronomical image denoising using curvelet and starlet transform

Abstract

Astronomical image data acquisition under low light conditions causes higher noise occurrence in these data. There are a lot of noise sources including also the thermally generated noise (dark current) inside used astronomical CCD sensor and the Poisson noise of the photon flux. There are specific image quality criteria in astronomy. These criteria are derived from the algorithms for astronomical image processing and are specific in the field of multimedia signal processing. Astrometric and photometric algorithms provide information about stellar objects: their brightness profile (PSF), position and magnitude. They could fail because of lower SNR. This problem can be solved by subtraction a dark frame from a captured image nowadays. However, this method couldn't work properly in systems with shorter shutter speed and nonlinear sensitivity, such as for example the system MAIA (Meteor Automatic Imager and Analyser). Image data from these system could not been processed by conventional algorithms. Denoising of the astronomical images is therefore still a big challenge for astronomers and people who process astronomical data. Therefore usage of other denoising algorithms is proposed in this paper. We describe our experiences with astronomical image data denoising based on Curvelet and Starlet transform. Novel algorithms have been tested on image data from MAIA system. Their influence on important photometric data like stellar magnitude and FWHM (Full Width at Half Maximum) has been studied and compared with conventional denoising methods.