Broadband Hyperspectral 3D Imaging using Dispersed Structured Light

2026-06-08 · Source: cs.CV updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Expert, extended

Summary

A new broadband hyperspectral 3D imaging (BH3D) method extends spectral and geometric capture across the full visible-near-infrared and short-wavelength infrared (SWIR) spectrum (450–1500 nm). This approach overcomes traditional limitations of narrow spectral windows and complex multi-spectrograph designs by employing a single-spectrograph system. It utilizes dispersed structured light with a stereo setup comprising visible and SWIR cameras to reconstruct dense broadband hyperspectral reflectance and accurate 3D geometry. The system achieves accurate reconstruction with a mean spectral angle mapper of 0.13 rad, root mean square error of 0.03, and mean depth error of 4.5 mm. Demonstrated applications include identifying metameric materials, imaging through opaque layers, uncovering hidden banknote features, and revealing blood vessels.

Key takeaway

For Computer Vision Engineers developing advanced material analysis or non-destructive inspection systems, this broadband hyperspectral 3D imaging method offers a unified solution. You can now capture comprehensive material properties and accurate 3D geometry from 450-1500 nm with a single spectrograph, simplifying hardware and calibration. Consider integrating this dispersed structured light approach to enhance your system's ability to identify subtle material differences or reveal hidden features.

Key insights

A single-spectrograph system enables broadband hyperspectral 3D imaging by extending dispersed structured light across visible-SWIR.

Principles

• Different wavelengths provide complementary physical cues.
• Broadband dispersed structured light unifies VNIR and SWIR observations.
• Stereo imaging resolves wavelength-geometry ambiguity.

Method

The system projects broadband spectrally encoded structured illumination using a single dispersive element and a galvo mirror, captured by VNIR/SWIR stereo cameras. An image formation model and optimization reconstruct hyperspectral reflectance and depth.

In practice

• Discriminate visually similar materials (e.g., baking soda vs. sugar).
• Inspect objects through opaque packaging films.
• Visualize subsurface biological structures like blood vessels.

Topics

• Hyperspectral Imaging
• 3D Imaging
• Structured Light
• SWIR Imaging
• Material Characterization
• Non-destructive Testing

Code references

• shshin1210/BH3D

Best for: AI Scientist, Computer Vision Engineer, Research Scientist

Related on AIssential

• Ambient-robust Inverse Rendering using Active RGB-NIR Imaging — Combining active NIR flash with multi-view RGB imaging yields ambient-robust inverse rendering for accurate geometry and reflectance.
• Voronoi-guided Bilateral 2D Gaussian Splatting for Arbitrary-Scale Hyperspectral Image Super-Resolution — GaussianHSI uses Voronoi-guided 2D Gaussian splatting for arbitrary-scale hyperspectral image super-resolution.
• Learning Sparse BRDF Measurement Samples from Image — Optimizing BRDF measurement locations with a differentiable renderer improves low-budget material reconstruction.
• Single Image Defogging Using a Fourth-Order Telegraph PDE Guided by Physical Haze Modeling — This paper introduces a novel hybrid defogging model that integrates a fourth-order nonlinear Telegraph Partial Differential Equation (PDE) with a physical haze formation model for single image defogging.
• RISE: Single Static Radar-based Indoor Scene Understanding — RISE utilizes multipath radar reflections and generative AI for privacy-preserving, geometry-aware indoor scene understanding from a single static sensor.
• DF3DV-1K: A Large-Scale Dataset and Benchmark for Distractor-Free Novel View Synthesis — DF3DV-1K is a new dataset for benchmarking distractor-free novel view synthesis methods.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.