Passive localization and reconstruction of multiple non-line-of-sight objects in a scene with a large visible transmissive window.

Abstract

Passive non-line-of-sight imaging methods have been demonstrated to be capable of reconstructing images of hidden objects. However, current passive non-line-of-sight imaging methods have performance limitations due to the requirements of an occluder and aliasing between multiple objects. In this paper, we propose a method for passive localization and reconstruction of multiple non-line-of-sight objects in a scene with a large visible transmissive window. The analysis of the transport matrix revealed that more redundant information is acquired in a scene with a window than that with an occluder, which makes the image reconstruction more difficult. We utilized the projection operator and residual theory to separate the reconstruction equation of multiple objects into the independent equations of the located objects that can be reconstructed independently by TVAL3 and Split-Bregman algorithms, which greatly reduces the computational complexity of the reconstruction. Our method lays the foundation for multiple objects reconstruction in complex non-line-of-sight scenes.