Research map / 3D vision / Depth and stereo
Active Hyperspectral Imaging Using an Event Camera
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes an event-camera-based active hyperspectral imaging system that encodes spectrum as temporal events under a sweeping-rainbow illumination and reconstructs spectra via constrained optimization.
Problem
Hyperspectral imaging faces a trade-off between capture speed, spectral resolution, and bandwidth, especially for dynamic scenes; existing scanning, snapshot, and coded-aperture methods cannot jointly achieve high temporal and spectral resolution with low bandwidth.
Approach
Uses a rotating mirror array and blazed grating to create a sweeping rainbow illumination so that wavelength maps to time; an event camera's log-intensity change events yield null spectrum vector constraints, solved with non-negativity and smoothness/anti-drift regularized QP.
Key results
- Achieves 59.53% bandwidth reduction versus frame-based systems with accuracy comparable to frame-based methods on static real scenes.
- On a 24-patch ColorChecker, reports average RMS error 0.204, SAM 21.05, and PSNR 28.46.
- Enables real-time hyperspectral video at 10 FPS (paper also claims up to 30 FPS in abstract/figure) for dynamic scenes.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use event-based spectral encoding instead of basis-based reconstruction | Active hyperspectral imaging · Rainbow pattern | SAM: 25.97 -> 23.27 | Helped |
| use event-based spectral encoding instead of basis-based reconstruction | Active hyperspectral imaging · Rainbow pattern | PSNR: 23.79 -> 24.61 | Helped |
| use event-based spectral encoding instead of bandwidth-matched frame capture | Active hyperspectral imaging · Rainbow pattern | PSNR: 18.73 -> 24.61 | Helped |
| add non-negative constraints (ablation) | Spectral reconstruction · Spectral reconstruction | SAM: 30.23 -> 13.42 | Helped |
| add non-negative constraints (ablation) | Spectral reconstruction · Spectral reconstruction | PSNR: 26.62 -> 33.60 | Helped |
| add low-frequency prior (ablation) | Spectral reconstruction · Spectral reconstruction | SAM: 30.97 -> 13.42 | Helped |
| add low-frequency prior (ablation) | Spectral reconstruction · Spectral reconstruction | PSNR: 26.42 -> 33.60 | Helped |
| add anti-drift constraints (ablation) | Spectral reconstruction · Spectral reconstruction | SAM: 26.28 -> 13.42 | Helped |
| add anti-drift constraints (ablation) | Spectral reconstruction · Spectral reconstruction | PSNR: 27.81 -> 33.60 | Helped |
Limitations
- Inter-reflection, scattering, and fluorescence are not modeled, affecting spectral accuracy.
- Requires precise optical calibration.
- Algorithm is not yet fast; implementation has room to speed up.
Experimental setup
Datasets: synthetic hyperspectral simulation data ColorChecker chart real metameric daisies
Benchmarks: ColorChecker chart metameric daisies high-frequency rainbow pattern real static and dynamic scenes
Metrics: RMS error SAM PSNR SSIM bandwidth reduction
Baselines: Full-bandwidth frame-based method Bandwidth-matched frame-based method Parkkinen basis method CASSI-based method
Builds on: Parkkinen basis method CASSI
Closest papers by meaning
- Event Fields: Capturing Light Fields at High Speed, Resolution, and Dynamic Range (CVPR 2025 · similarity 0.86)
- SPECAT: SPatial-spEctral Cumulative-Attention Transformer for High-Resolution Hyperspectral Image Reconstruction (CVPR 2024 · similarity 0.84)
- Dense Dispersed Structured Light for Hyperspectral 3D Imaging of Dynamic Scenes (CVPR 2025 · similarity 0.84)
- Dispersed Structured Light for Hyperspectral 3D Imaging (CVPR 2024 · similarity 0.83)
- Spectral Compressive Imaging via Chromaticity-Intensity Decomposition (NeurIPS 2025 · similarity 0.82)
- FRN: Fractal-Based Recursive Spectral Reconstruction Network (NeurIPS 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.