Ev2Gray: Generating Intensity Images from Event Streams

Institute for Industrial Information Technology (IIIT), Karlsruhe Institute for Technology (KIT)

johannes.bassler@kit.edu

Abstract

Event-based vision sensors surpass traditional frame-based cameras for capturing dynamic changes, yet they struggle with static regions and lack absolute intensity information, limiting scene understanding. This work addresses that by generating absolute intensity images without a secondary camera. The key idea is a moving opaque strip that traverses the field of view, deliberately occluding and then revealing parts of the scene. Since the occlusion intensity is identical for all pixels, it provides a universal reference point. By counting positive and negative events during occlusion and revelation, the method yields a relative intensity value with respect to the occlusion-level brightness — even in otherwise static regions. By producing meaningful grayscale images while preserving the advantages of event-based sensors and introducing minimal disturbance through brief, localized occlusion, our method bridges the gap between relative event output and absolute intensity information, thereby enabling broader applications for event-based vision systems. 78 ebungstechnologien und Spezialkameras R. Kowarschik F R E I T A G

Manuskript noch nicht eingereicht. Der Vortragende kann unter /einreichen mit Code (A33) und der hinterlegten E-Mail-Adresse einen Upload-Link anfordern.
@inproceedings{dgao126-a33, title = {Ev2Gray: Generating Intensity Images from Event Streams}, author = {J. Baßler, M. Heizmann}, booktitle = {DGaO-Proceedings, 126. Jahrestagung}, year = {2025}, publisher = {Deutsche Gesellschaft für angewandte Optik e.V.}, issn = {1614-8436}, note = {Talk A33} }
126. Annual Conference of the DGaO · Stuttgart · 2025