Dynamic Reconstruction from Neuromorphic Data

TL;DR

This work addresses reconstructing frame-like images and scene dynamics from purely neuromorphic event data, circumventing the need for conventional camera frames. It models the per-pixel luminosity evolution as an initial-value problem driven by temporally-derivative information derived from events, and solves a regularized least-squares problem to recover the temporal sequence of pixel intensities $v_{xy}(t)$ using only event data. The approach provides a physical interpretation via an ODE-like framework and demonstrated efficacy on real datasets, including handheld office scenes, ballistic cup dynamics, and ISS Falcon imagery, with regularization parameter $oldsymbololdsymbol\lambda$ controlling the balance between data fidelity and temporal smoothness. This method advances practical event-based reconstruction by removing dependence on synchronized traditional frames, enabling robust, high-temporal-resolution imaging in dynamic real-world environments.

Abstract

Unlike traditional cameras which synchronously register pixel intensity, neuromorphic sensors only register `changes' at pixels where a change is occurring asynchronously. This enables neuromorphic sensors to sample at a micro-second level and efficiently capture the dynamics. Since, only sequences of asynchronous event changes are recorded rather than brightness intensities over time, many traditional image processing techniques cannot be directly applied. Furthermore, existing approaches, including the ones recently introduced by the authors, use traditional images combined with neuromorphic event data to carry out reconstructions. The aim of this work is introduce an optimization based approach to reconstruct images and dynamics only from the neuromoprhic event data without any additional knowledge of the events. Each pixel is modeled temporally. The experimental results on real data highlight the efficacy of the presented approach, paving the way for efficient and accurate processing of neuromorphic sensor data in real-world applications.