First-photon imaging with independent depth reconstruction

Abstract

First-photon imaging allows the reconstruction of scene reflectivity and depth information with a much fewer number of photon countings, compared with conventional time-correlated single-photon counting based imaging systems. One problem of the original first-photon imaging is that the quality of depth reconstruction is significantly based on the denoising effect led by the result of reflectivity reconstruction; therefore, once the detection environment has a low SBR (signal-to-background ratio), the depth image denoising and reconstruction result is poor. In this work, an improved first-photon imaging scheme is proposed, in which the depth is reconstructed independently by optimizing the denoising method. A denoising module based on K-singular value decomposition is applied to remove the practical noise, including ambient noise and the dark count of the detector before the reconstruction of the depth image. The numerical and experimental results demonstrate that the proposed scheme is capable of denoising adaptively under different noise environments, especially severe ones. Under the condition of SBR being 1.0, the averaged root mean square error of depth reconstruction images is 36.2% smaller than that of the original first-photon imaging scheme.