Super-resolution in imaging through a transparent spherical microlens has attracted lots of attention because of recent promising experimental results with remarkable resolution improvement. To provide physical insight for this super-resolution phenomenon, previous studies adopted a phenomenological explanation mainly based on the super-focusing effect of a photonic nanojet, while a direct imaging calculation with classical imaging theory has rarely been studied. Here we theoretically model the imaging process through a microlens with vectorial electromagnetic analysis, and then exclude the previously plausible explanation of super-resolution based on the super-focusing effect. The results showed that, in the context of classical imaging theory subject to the two-point resolution criterion, a microlens with a perfect spherical shape cannot achieve the experimentally verified sub-100 nm resolution. Therefore, there must be some other physical mechanisms that contribute to the reported ultrahigh resolution but have not been revealed in theory.