Per-pixel calibrated event-domain inversion for slanted-edge modulation transfer function estimation

Abstract

Event cameras operate asynchronously and provide high dynamic range with microsecond temporal resolution. However, their output does not correspond to conventional grayscale intensity measurements, precluding the direct application of standard electro-optical metrics such as slanted-edge Modulation Transfer Function (MTF). This work proposes a methodology to estimate MTF by deriving intensity-equivalent representations from event data, using both count-based reconstruction and latency-derived mappings, followed by conventional edge analysis. A key aspect of the approach is a per-pixel calibration, where a controlled, dynamically modulated LED stimulus is used to map pixel-level response behavior across rate and temporal domains. These mappings are applied to slanted-edge targets to reconstruct intensity-equivalent images suitable for classical MTF analysis. A beam-splitter setup provides a shared optical path with a matched frame camera for comparison under controlled defocus. Preliminary results show consistent defocus trends across both systems and indicate that per-pixel calibration improves the stability of event-based MTF estimation.