An inverse problem resolution for 3D particle tracking using event-based vision

Overview & Synopsis

Particle tracking in fluid systems is fundamental for the characterization of complex flow dynamics, but current methods are constrained by temporal and spatial resolution of conventional frame-based cameras. This paper proposes the application of the emerging event-based technology to reconstruct a 3D trajectory from 2D projections acquired by event-based cameras. Building on the pinhole camera model, we reformulate the well-established relation between the 2D projections and 3D estimate. This allows us to define a convex data-fidelity term, and a reconstruction procedure based on the minimization of a penalized least-squares functional. Experimental results show that this approach enables the reconstruction of a 3D trajectory with high accuracy. This method opens new perspectives for the analysis of complex flows in fluid systems.