Reproduce “Benchmarking Image Sensors Under Adverse Weather Conditions for Autonomous Driving” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mario Bijelic, Tobias Gruber, Werner Ritter
This paper establishes a comprehensive methodology for benchmarking automotive image sensors under controlled fog conditions, demonstrating that a near-infrared (NIR) gated camera significantly outperforms a standard CMOS camera in maintaining image contrast and mitigating backscatter in dense fog, even with challenging active light sources. The approach utilizes a climate chamber for repeatable and quantitative evaluation of emerging sensor technologies.
Key insights
- Standard passive cameras suffer significant contrast degeneration in fog due to light attenuation and backscatter ('air-light'), which obscures objects and reduces image information content.
- Gated imaging, utilizing time-synchronized active NIR illumination and short exposure times, effectively mitigates backscatter and maintains higher contrast and information content in foggy environments.
- The proposed methodology, including an adapted scattering model, successfully provides a quantitative basis for understanding sensor behavior in fog and allows for objective benchmarking and potential extrapolation of results.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.