Experimental demonstration of imaging hidden objects in opaque liquid-based media by fusion of single-shot multiview polarized and unpolarized speckle images

Abstract Turbid water-based liquids are common scattering media within which imaging objects is challenging, since scattering particles of various origins and sizes may cause severe image degradation, resulting in poor object recovery and resolution. Therefore, counteracting the nontrivial effects of light scattering is a primary challenge, limiting the utility of optical imaging within liquid media. In this work, we sought to circumvent the optical degradation in turbid liquid media and to recover high-quality images of hidden objects by averaging polarized speckle images projected from several directions through a lens array. Inspired by astronomy imaging techniques, the averaging process utilized a Shift-and-Add (SAA) algorithm developed to reconstruct atmospherically-degraded solar images. Both linear and circular polarization strategies were then applied to further improve image reconstruction. Experiments were carried out using two targets, each having a distinct geometrical shape, embedded within turbid liquid media comprising a range of dilutions of commercially available cow's milk in seawater. The media were irradiated with a polarized laser beam, after which multiple polarized speckled images were captured with a CCD camera. Offline, images were first shifted to a common center and then fused, using the SAA algorithm, to yield reconstructed images of each hidden object. Quantitative image quality metrics, including SNR, entropy, and sharpness were applied to evaluate quality of the reconstructed images. The reconstructed images obtained demonstrate successful adaptation of our methodology for the identification of hidden objects within turbid media. To the best of our knowledge, this is the first report to demonstrate successful imaging of hidden object images within turbid liquid media by averaging multiple polarized speckle projections.