High-fidelity correspondence imaging in complex media with varying thresholds and 1-bit compressive sensing

Abstract

Here, we report high-fidelity correspondence imaging (CI) in complex media. By introducing varying thresholds to binarize single-pixel light intensities recorded in complex media, dynamic scaling factors can be eliminated. Then, the binarized light intensities and illumination patterns can be fed into a modified 1-bit compressive sensing algorithm to realize high-fidelity object reconstruction. The proposed method can implement object reconstruction with high fidelity in complex media without extra temporal carriers. It is experimentally verified that the method can effectively eliminate dynamic scaling factors and realize high-fidelity object reconstruction in complex media where conventional CI methods could fail. Experimental results demonstrate that the proposed method broadens a potential application of CI in complex media, e.g., turbid water, biological tissues, and dynamic smoke.