Steady-state Non-Line-of-Sight Imaging.

Conventional intensity cameras recover objects in the direct line-of-sight of the camera, while occluded scene parts are considered lost in the image formation process. Non-line-of-sight imaging (NLOS) aims at recovering these occluded objects by analyzing their indirect reflections on visible scene surfaces. Existing NLOS methods temporally probe the indirect light transport to unmix light paths based on their travel time, which mandates instrumentation with low photon efficiency, high cost, and mechanical scanning. We depart from temporal probing and demonstrate steady-state non-line-of-sight imaging using conventional intensity sensors and continuous illumination. Instead of assuming perfectly isotropic scattering, the proposed method exploits directionality of reflectance on object surfaces resulting in spatial variation of the indirect reflections for varying illumination patterns. To tackle the shape-dependence of these variations, we propose a trainable architecture which learns from synthetic data to map steady-state diffuse indirect reflections to unknown scene reflectance. Relying on consumer color image sensors, with high fill-factor, quantum efficiency and low read-out noise, we demonstrate high-fidelity full-color NLOS imaging in setup scenarios identical to recent pulsed systems with picosecond resolution.