EUVIMEDEUVIMED entdecken
Crossref

Characterizing, Modeling, and Mitigating Radiation-Induced Noise on Event-Based Vision Sensors for Space Applications

Joshua B. Poravanthattil, Seth Roffe, Alan D. George

IEEE Sensors Journal · 2026-11 · 26 (21) · 32464-32478

DOI: 10.1109/jsen.2026.3734852 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, 2026-11. Verfasst von Joshua B. Poravanthattil und 2 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Gefördert von: NSF Center for Space, High-Performance, and Resilient Computing (SHREC) Industry and Agency Members, Industry-University Cooperative Research Centers (IUCRC) Program of the National Science Foundation, Los Alamos Neutron Science Center (LANSCE), a National Nuclear Security Administration (NNSA) User Facility Operated for the U.S. Department of Energy (DOE) by….

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Event-based vision sensors offer several key advantages for aerospace and autonomous systems, including high temporal resolution, low latency, and efficient data representation. However, their susceptibility to radiation-induced noise remains an important concern for deployment in space environments. This research investigates the effects of radiation on event-based sensors and evaluates statistical modeling, fault injection, and event filtering approaches for characterizing and mitigating radiation-induced activity. Experimental data were collected under neutron irradiation using two representative sensors, the Prophesee EVK1 and the iniVation DVXplorer Lite, with the lens caps attached to isolate intrinsic sensor activity and radiation-induced events. A statistical analysis of windowed event-rate dynamics was performed using metrics including mean event rate, coefficient of variation, and peak-to-mean ratio to characterize the temporal structure of the resulting event streams. The results reveal substantial differences in radiation response between the two sensors. The Prophesee EVK1 exhibits large increases in burst-dominated activity, while the DVXplorer Lite produces a more uniform increase in event activity that more closely resembles intrinsic background noise. These structural differences significantly influence filtering performance, with conventional event-denoising algorithms demonstrating high suppression efficacy for the DVXplorer Lite but reduced effectiveness for the Prophesee EVK1. In addition, this research enhances the Event-based Radiation-Induced Noise Simulation Environment (Event-RINSE) using independently fit, polarity-specific empirical models for both sensors. Held-out comparisons between measured and generated responses quantitatively evaluate the temporal fidelity of the resulting fault models. An application-level evaluation using E-RAFT and labeled DSEC optical-flow data further demonstrates rate-dependent performance degradation under EVK1-model injection, while the lower magnitude DVXplorer Lite model produces no practically meaningful degradation over the evaluated interaction-rate range. The results highlight the importance of considering sensor-specific radiation dynamics, realized injected-event burden, and downstream algorithm sensitivity when selecting event-based sensors and designing resilience strategies for space perception systems.

Förderung

  • NSF Center for Space, High-Performance, and Resilient Computing (SHREC) Industry and Agency Members
  • Industry-University Cooperative Research Centers (IUCRC) Program of the National Science Foundation
  • Los Alamos Neutron Science Center (LANSCE), a National Nuclear Security Administration (NNSA) User Facility Operated for the U.S. Department of Energy (DOE) by…

Zitieren

APA

Poravanthattil, J. B., Roffe, S., & George, A. D. (2026). Characterizing, Modeling, and Mitigating Radiation-Induced Noise on Event-Based Vision Sensors for Space Applications. IEEE Sensors Journal, 26(21), 32464-32478. https://doi.org/10.1109/jsen.2026.3734852

Vancouver

Poravanthattil JB, Roffe S, George AD. Characterizing, Modeling, and Mitigating Radiation-Induced Noise on Event-Based Vision Sensors for Space Applications. IEEE Sensors Journal. 2026;26(21):32464-32478. doi:10.1109/jsen.2026.3734852

BibTeX
@article{Poravanthattil2026,
  author = {Poravanthattil, Joshua B. and Roffe, Seth and George, Alan D.},
  title = {Characterizing, Modeling, and Mitigating Radiation-Induced Noise on Event-Based Vision Sensors for Space Applications},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {21},
  pages = {32464--32478},
  doi = {10.1109/jsen.2026.3734852}
}
Kostenlos starten →