Computational Nonscanning Incoherent Superoscillatory Imaging

Superoscillatory (SO) imaging is an emerging technique to super-resolve unlabeled objects in the far-field. Reducing the full width at half-maximum (FWHM) of the main beam is a standard method used in SO imaging to pursue a finer resolution. However, reducing the FWHM dims the main beam sharply. This results in very poor signal-to-noise ratios that are beyond the capability of conventional image sensors. We present an approach that does not seek to reduce the FWHM of the main beam. Instead, we describe an imaging system whose SO point spread function is very broad, yet preserves sufficiently high frequencies to enable sharp image reconstruction by computational deconvolution. A key observation in this work is that deconvolution-based SO imaging is only possible for SO systems that are incoherent; we show how to realize such a system with a red light-emitting diode and a programmable spatial light modulator. This system enables the application of standard deconvolution algorithms to image subdiffraction objects in a single shot without any form of scanning. Overall, we demonstrate computational SO imaging of previously unseen 2D complex objects with a submicron resolution that is one-fifth the diffraction limit.