Current research in life sciences heavily relies on optical microscopy. Correctly characterizing the involved systems is crucial since it directly impacts the final imaging performance, in particular after deconvolution. Typical parameters (magnification, numerical aperture, wavelength, refraction indices, etc.) greatly depend on the actual experimental conditions, including the objective, the filter set, the immersion media, temperature and pH. We propose to estimate these optical parameters by acquiring an image stack of reference objects such as beads or slits, and fitting a PSF model to the data. The project will comprise three main tasks: 3D modeling of the imaging process, MATLAB implementation and experimental validation on a Zeiss Axioskop 2 microscope. In particular, we aim at using the obtained parameters for deconvolution.