A method to analyze adaptive optics images of binary stars

We present a fast and robust ad-hoc method to analyze a set of short exposure images of a close bi- nary star taken by the adaptive optics system of the ESO 3.6 m telescope. At rst each image is tted, assuming as a radially symmetrical approximation of its \instan- taneous" PSF, the superposition of a central Gaussian and of a Moat torus. Owing to the coral variability and asymmetry of the PSF the t gives a poor estimate of the component's luminosity ratio; nevertheless it reproduces satisfactorily the global structure of the PSF and above all gives the accurate sub-pixel positions of the two stars. This allows to obtain, by the shift-and-add procedure, the mean set image from which the nal parameters of the two stars and the mean PSF during the observing run are derived. The results from a set of AO images of the close binary system Canis Maioris are consistent with those previously obtained by one of us using the iterative blind deconvolution. Extensive simulations give evidence that the magnitude dierence and the separation of the components can be estimated within about 0.05 mag and 0: 005 respectively when the secondary star image does not overlap a bright artifact of the AO system.