Compressed imaging with focused light

Compressed sensing applied to optical microscopy enables imaging with a number of measurements below the Nyquist criterion. The illumination basis selected, often unstructured for randomness considerations, influences the performance of image reconstruction algorithms. Here, we show through modelling based on multimode fiber imaging that an illumination basis composed of a series of uniformly spaced foci provides improved robustness to noise, increased volumetric imaging performance, and greater resilience to external perturbation when compared to a speckle illumination basis. These observations have broad implications for computational super-resolution imaging, endo-microscopy, and post-processing of images acquired with any point-scanning imaging system.