Improving the signal-to-noise ratio in incoherent imaging

Imaging plays an important role in our modern world. Besides limitations in terms of resolution, all imaging systems show degrading image quality with increasing noise. Hence, an important aspect of any imaging application is the achievable signal-to-noise ratio (SNR). When neglecting typical noise sources (e.g. from detectors), we are left with a very fundamental noise type: shot noise. Leading to a noisy measurement process which fundamentally limits the achievable SNR. In this thesis that limitation will be overcome by employing an approach termed splitting & recombination. The basic idea is to actively influence a conventional imaging system in such a way, that different "views" from the sample are being captured, which then get computationally recombined to yield an effective increase in SNR. Four possibilities for the splitting are being examined: 1) separating the polarization of the detectable light into two orthogonal states; 2) splitting the emission in a fluorescence microscope; 3) splitting the detection pupil; 4) separating the illumination in a light-sheet microscope using the Field-Synthesis concept. The two main computational recombination techniques employed are: weighted averaging in Fourier space and multiview deconvolution. For the aforementioned applications, a thorough theoretical and numerical treatment is presented, as well as proof-of-principle experiments. The second part of this thesis focuses on image reconstruction techniques of data acquired with a passive submillimeter wave imager, which has been developed for concealed threat detection in public areas. Pre-processing is necessary to transform the recorded data into meaningful images. These are then further processed via denoising and deconvolution algorithms to reveal small structures with improved SNR, so that a reliable detection of unconcealed threats becomes feasible. The robustness of all reconstruction techniques is shown, by analyzing a multitude of experimental data/scenarios.