The procedure derives first a PSF digital template from the brightest isolated field stars; then a catalogue of presumed objects is formed, searching for the relative intensity maxima in the image frame. In the following step the images of the suspected stars are analyzed in order of decreasing luminosity. In this phase a catalogue including the accepted objects is formed and a synthetic image of the observed field is constructed, placing an intensity-scaled PSF template in the position of each identified star. Each suspected object in the original list is accepted on the basis of its correlation coefficient with the PSF template; the relative astrometry and photometry of the source are determined by means of a fit, taking into account the contribution of the local non-uniform background and of the already detected stars, known from the synthetic field. As the analysis proceeds, fainter and fainter sources can be successfully analyzed, discriminating their central peaks from the secondary bumps of light in the distorted diffraction rings of the neighboring more luminous already identified stars: in this way the synthetic field becomes more and more similar to the observed image. Residual unexplained features of the image may be further analyzed checking for indications of blended groups, at separations smaller than the PSF FWHM.
StarFinder should be regarded as a tool for high precision astrometry and photometry of crowded fields acquired under the above assumptions of accurate PSF knowledge, isoplanatism and correct sampling. An application of StarFinder to HST undersampled data handled by dithering techniques (Fruchter et al. 1997) has been shown in Aloisi et al. (2000). In this paper it can be seen that the results obtained by our method are comparable to those obtained by DAOPHOT (Stetson 1987).
As far as the PSF is concerned, the accurate knowledge of its features outside the central peak is fundamental to perform a deep study of a stellar field, achieving accurate photometry of faint stars and avoiding either false detections or star loss (Esslinger & Edmunds 1998). If the PSF template cannot be extracted directly from the field, due to extreme crowding or lack of bright isolated stars, StarFinder can still be applied using a PSF estimated by means of other methods, as the reconstruction technique proposed by Véran et al. (1997) for AO observations or the TINY TIM simulation software for HST (Krist & Hook 1999).
Much more intriguing and difficult to solve is the case of a field with space variant PSF, either due to anisoplanatic effects as in AO observations or to design and control in HST images. In general the analysis of an anisoplanatic field requires the knowledge of the local PSF. A method to reconstruct the off-axis PSF in AO imaging has been proposed by Fusco et al. (2000). StarFinder, in its present version, can analyze frames with space invariant PSF or sub-frames in which the isoplanicity condition is nearly satisfied, as will be shown in Sect. 4; the extension to the space variant case is in progress, with preliminary results presented by Diolaiti et al. (2000a, 2000b).
The paper is organized as follows: the general features of the algorithm are described in Sect. 2; Sect. 3 deals with more technical aspects and might be skipped on a first reading; the method is validated on simulated and experimental data in Sect. 4; details on the IDL code are presented in Sect. 5; Sect. 6 includes our conclusions and future plans.
Copyright The European Southern Observatory (ESO)