This paper presents a Bayesian framework for non-line-of-sight (NLOS) imaging that overcomes limitations of existing methods by eliminating the need for dense measurements at regular grid points on the relay surface. A novel Confocal Complemented Signal-Object Collaborative Regularization (CC-SOCR) algorithm is introduced, leveraging virtual confocal signals to enable high-quality reconstructions of both albedo and surface normal, even with coarse or irregular measurement patterns. This approach significantly expands the applicability of NLOS imaging to various real-world scenarios.
