Bilevel Inverse Problems in Neuromorphic Imaging

TL;DR

This paper introduces a bilevel inverse problem framework for neuromorphic imaging, providing theoretical guarantees and a Newton-type solver, demonstrated through several examples, to improve processing of asynchronous event data.

Contribution

It develops a novel bilevel inverse problem formulation for neuromorphic cameras, including existence proofs, second order conditions, and a solver, advancing computational methods in this field.

Findings

Existence of solutions to the inverse problem is established.

A second order Newton type solver is derived and validated.

The approach demonstrates effectiveness on multiple example scenarios.

Abstract

Event or Neuromorphic cameras are novel biologically inspired sensors that record data based on the change in light intensity at each pixel asynchronously. They have a temporal resolution of microseconds. This is useful for scenes with fast moving objects that can cause motion blur in traditional cameras, which record the average light intensity over an exposure time for each pixel synchronously. This paper presents a bilevel inverse problem framework for neuromorphic imaging. Existence of solution to the inverse problem is established. Second order sufficient conditions are derived under special situations for this nonconvex problem. A second order Newton type solver is derived to solve the problem. The efficacy of the approach is shown on several examples.