Non-line-of-sight imaging with arbitrary illumination and detection pattern

Non-line-of-sight (NLOS) imaging aims at reconstructing targets obscured from the direct line of sight. Existing NLOS imaging algorithms require dense measurements at regular grid points over a large relay surface, limiting applicability in practical scenarios (robotic vision, autonomous driving, rescue, remote sensing). This work proposes a Bayesian framework for NLOS imaging without constraints on the spatial pattern of illumination and detection points. By introducing virtual confocal signals, we develop a confocal complemented signal-object collaborative regularization (CC-SOCR) algorithm enabling high-quality reconstructions. The approach reconstructs both albedo and surface normals with fine details under general relay settings. With a regular relay surface, coarse rather than dense measurements suffice, significantly reducing acquisition time. Experiments demonstrate that this framework substantially extends the application range of NLOS imaging.