Abstract

Latency and noise are crucial aspects of Event-based Vision Sensors. Yet, in simulators used to create synthetic event data these effects are predominantly modeled phenomenologically and are rarely calibrated to actual measurements or circuit simulations. This work presents a physics-based latency and noise model achieving strong resemblance with circuit simulations and measurements. The model is computationally efficient enough to be suitable for camera simulation. This enables accurate training data synthesis for algorithm development and guides sensor design.