A.K.A. How to deal with ambient light when using SPAD as a 3D camera
Most use cases of SPAD as a 3D camera comes with an inherent problem: ambient light. The 10m-10km measurement range of SPAD cameras means that most uses will be outdoors where ambient light is difficult to control. This paper tackles that by averaging out the ambient light. In contrast to the time of flight graph of a photon to the actual measurement target, which appears as a single high frequency mode in the time-intensity graph, ambient light is a flat low frequency line that stays constant. Because a single SPAD diode can only detect a single photon every so often, the ambient light has a high chance of hitting the diodes and leads to “pileup”. This paper fixes the issue by differing the timestamp at which a laser pulse transmits and one at which a SPAD sensor starts acquiring photons. This way, the pile of ambient photons appear at different locations in the graph relative to the point of interest, and they can be averaged out.
Problem: inconsistent ambient light
While the authors’ model of computational synchronization is simple and elegant, it relies on the assumption that ambient light stays constant. This can be a problem if the camera, object, or light source is moving around, or there are other light sources other than a constant ambient light. Perhaps SPAD works at a high enough frequency to work under these conditions too. Combining SPAD based 3D cameras with Epiolar ToF methods may be able to mitigate ambient light better.