Event-based camera characterization

Unlike conventional area-scan cameras, event-based cameras lack standardized characterization using application-oriented quality metrics that concisely capture fundamental properties with minimal parameters. Consequently, direct comparisons between these camera types—or efficient selection for specific applications—are currently infeasible. This work adopts the systems-theoretic foundation of the EMVA 1288 standard, which enables camera characterization via photon flux as input and digital values as output signal, without requiring camera’s internal operational knowledge. However, the standard’s time-constant, spatially homogeneous stimuli are incompatible with event-based sensors. Accordingly, pulsed stimuli with variable background illumination and adjustable contrast were employed, therefore the characterization is performed with respect to irradiance and not exposure. Three novel, universal quality metrics were developed—Event Contrast Detectability (ECD), latency, and Event Detection Efficiency (EDE)—applicable across sensor types. By emulating event-based behavior with standard area-scan cameras, these metricswere analytically linked to signal-to-noise ratio and frame rate. Single-photon avalanchediode (SPAD) sensors with configurable bit depth were integrated into the comparativeframework. The practical utility of these metrics was validated through measurementson three event-based cameras, three area-scan cameras, and two SPAD cameras. Finally,optimal cameras, illumination strategies, and configurations were identified for threerepresentative industrial tasks: object sorting, motion tracking, and vibration monitoring.