Adaptive Optics (AO) images are characterized by structured Point Spread Function (PSF), with sharp core and extended halo, and by significant variations across the field of view. In order to enable the extraction of high-precision quantitative information and improve the scientific exploitation of AO data, efforts in the PSF modeling and in the integration of suitable models in a code for image analysis are needed. We present the current status of a study on the modeling of AO PSFs based on observational data taken with present telescopes (VLT and LBT). The methods under development include parametric models and hybrid (i.e. analytical / numerical) models adapted to various types of PSFs that can show up in AO images. The specific features of AO data, such as the mainly radial variation of the PSF with respect to the guide star position in single-reference AO, are taken into account as much as possible. The final objective of this project is the development of a flexible software package, based on the Starfinder code (Diolaiati et Al 2000), specifically dedicated to the PSF estimation and to the astrometric and photometric analysis of AO images with complex and spatially variable PSF.