Optical linear systems framework for event sensing and computational neuromorphic imaging

Event Vision Sensors, or neuromorphic cameras, report sparse, and asynchronous image change-related data and enable microsecond-scale sensing and high dynamic range, but challenge physics-based sensor design approaches. In response to log-intensity threshold-crossing instances, this event representation does not readily integrate with forward operators used to describe most computational imaging and optical systems. We present a physics-grounded processing pipeline that maps event streams to estimates of per-pixel log-intensity and its derivatives, and embeds these measurements in a dynamic linear systems model with a time-varying Point-Spread Function (PSF). This enables inverse filtering directly from event data via frequency-domain Wiener deconvolution with a known dynamic transfer function, enabling scene inference even when no image is formed on the sensor plane. We validated the approach in simulation for single and overlapping point-sources under modulated defocus, and on real event data from a tunable-focus telescope imaging a star field, demonstrating superior source localization and separability compared with frame-based imaging. The proposed framework provides a practical bridge between event sensing and model-based computational imaging for dynamic optical systems.