Diffraction imposes for each optical system a resolution limit which could be described by using the vectorial theory of Richards and Wolf. This theory defines the intensity distribution of a point like source imaged by a lens assuming ideal imaging conditions. Unfortunately, these conditions can not be completely achieved in practical situations as a recorded microscope image is always affected by noise which makes the resolution limit worse. In this work we propose and analyze optical set-up schemes towards an image quality improvement in terms of Signal to Noise Ratio (SNR) in linear and non-linear fluorescence microscopy. In order to reach this purpose we insert, on the illumination arm of the microscope, a proper amplitude ring filter inducing laterally interfering beams. The effect induced by the filter results in a shape engineering of the 3D-PSF and in a redistribution of the spatial frequencies of the OTF. In particular, the high frequencies information are collected at improved SNR. In order to implement such schemes we use a computational simulation mainly based on a vectorial approach analyzing the results in both space and frequency domain to characterize the optical system response. Analysis reveals that, although the theoretical resolution of the system is unchanged, when we impose a certain noise level the practical imaging quality could be improved in the ring filtering scheme. The results suggest that further improvement can be reached by the usage of the proposed annular filers in combination with image restoration. A comparison between linear and non-linear excitation cases is presented.