Non-invasive imaging through thin scattering layers with broadband illumination

Abstract: Memory-effect-based methods have been demonstrated to be feasible to observe hidden objects through thin scattering layers, even from a single-shot speckle pattern. However, most of the existing methods are performed with narrowband illumination or require point light-sources adjacent to the hidden objects as the references, to make an invasive pre-calibration of the imaging system. Here, inspired by the shift-and-add algorithm, we propose that by randomly selecting and averaging different sub-regions of the speckle patterns, an image pattern resembling the autocorrelation (we call it R-autocorrelation) of the hidden object can be extracted. By performing numerical simulations and experiments, we demonstrate that comparing with true autocorrelation, the pattern of R-autocorrelation has a significantly lower background and higher contrast, which enables better reconstructions of hidden objects, especially in the case of broadband illumination, or even with white-light.