Calibration of Event-Based Camera for High-Speed Observations of Lightning and Transient Luminous Events

Abstract

Event-based vision sensors (EVSs) offer an innovative approach to optical sensing by responding solely to changes in photon flux with asynchronous pixels. This enables high-speed imaging with minimal data throughput and a large dynamical range > 120 dB. Despite their growing use in edge detection and the observation of Earth science phenomena such as lightning and transient luminous events (TLEs), optimizing EVS performance specifically for high-speed observations on microsecond timescales, where the photon flux is key, remains challenging. This is due to the unitless, non-intuitive nature of their bias settings, which drives the working principles of the sensor. In this study, we perform a comprehensive calibration of an event-based camera, focusing on three critical parameters: contrast thresholds, refractory period, and corner frequency of the inherent low-pass filter. Using a pre-calibrated LED source and newly developed event generation code, we establish physical relationships between these raw digital bias settings and the camera’s response characteristics. Our results show tunable ranges of 5–190% for on contrast thresholds, 5–65% for off contrast thresholds, 12–480 μs for refractory periods, and 1–11 kHz for corner frequencies. These calibrations provide the fundamental camera characteristics for accurate optical signal reconstruction and improved detection of fast transient phenomena. The approach is demonstrated on a Prophesee EVK4; however, the proposed calibration method can be applied to any EVS camera that employs similar hardware components and bias settings and thus provides a necessary framework for the quantitative use of EVS cameras in atmospheric remote sensing.