3D Scene Estimation with Perturbation-Modulated Light and Distributed Sensors

In this paper, we present a framework to roughly reconstruct the 3D occupancy scenario of an indoor space using color-controllable light and distributed color sensors. By applying randomly generated perturbation patterns onto the input of the LED fixtures, and measuring the changes of the sensor readings, we are able to recover the light transport model (LTM) of the room. Then a variant of the inverse Radon transform is applied on the LTM to reconstruct the 3D scene. The reconstructed scene by our algorithm can faithfully reveal the occupancy scenario of the indoor space, while preserving the privacy of human subjects. An occupancy-sensitive lighting system can be designed based on this technique.