Microscopic imaging has a resolution that is often far from the diffraction limit due to aberrations induced by the optics or by the sample itself. It is therefore of interest sensing these aberrations either directly or indirectly to improve image quality in post-processing or with adaptive optics. To avoid the use of extra hardware, several techniques are available to algorithmically retrieve aberrations in microscopy, but they often require images of an isolated point source, which is seldom a reasonable assumption as even single fluorescent beads often have a non negligible size. A different technique, known in literature joint estimation or phase diversity, was first developed by Gonsalves to estimate phase aberrations when imaging extended objects. In this master thesis we will derive a simple physical model that accounts for the vectorial nature of light and use it to expand the algorithms derived from Gonsalves seminal papers. We will show with numerical simulation that the effect of polarization is indeed non negligible when imaging through high-NA lenses, and therefore a vectorial model can substantially decrease model deviation and improve the quality of the estimates. The novel algorithm is thoroughly tested both in simulation and with experimental data.