From 3D to 5D: Computational light-in-flight imaging via an SPAD camera

Light-in-flight imaging, which images light itself at high frame rates, is an intuitive method for studying light-related ultrafast phenomena. Usually, light propagation happens in five dimensions, i.e. (x, y, z, t, n) if the refractive index of the medium is unknown. However, state-of-art imaging devices, such as single-photon avalanche diode cameras, capture only three-dimensional (x′, y′, t′) data, which are merely the projection of the five-dimensional information on the light-sensitive detector array. The recovery of the five-dimensional information from its three-dimensional projection is an intrinsically ill-posed problem. By considering the continuous nature of light propagation, this work demonstrates that it is possible to retrieve the five-dimensional information from the captured data of a single-photon avalanche diode camera. The feasibility of the proposed method is verified in different scenarios, including light propagating in a uniform medium, travelling along curved optical fibers, and bending in a gradient refractive index medium, with an averaged reconstruction accuracy of about 1 %. The proposed method broadens the applicable scope of light-in-flight imaging and provides insights for high-dimensional information retrieval.