Single-shot non-line-of-sight imaging based on the statistical average characteristics of a speckle pattern under ambient light

Abstract

To address the challenges of a small field of view (FOV) and multiple measurements in non-line-of-sight (NLOS) imaging, we propose a method for reconstructing the target object based solely on a random speckle pattern captured by a camera with coherent illumination. Moreover, the imaging FOV is not limited by optical memory effect (OME). Our method is based on the statistical average characteristics of a speckle pattern, and the amplitude information of the object spectrum is extracted through the covariance of this speckle pattern to reconstruct the object. Experimental results demonstrate that our method reconstructed different objects of 1.2 cm in size which are a 2.4-fold OME range, at a distance of 45 cm from the highly scattering wall, with a resolution of approximately 225 $\mu m$. Furthermore, to address the issue of ambient light damaging the effective information in the speckle patterns, by combining with the first 21 Zernike polynomials, the object was reconstructed under low signal-to-noise ratio (SNR) of −2.06 dB, promoting the application of our method in practical scenarios.