Broadband Hyperspectral 3D Imaging using Dispersed Structured Light

Researchers have demonstrated a broadband hyperspectral 3D imaging method that captures scene geometry and spectral data from 450 to 1500 nanometers using a single spectrograph and a stereo camera pair, according to a paper posted to arXiv on 25 May 2026 [1]. The approach, termed broadband hyperspectral 3D imaging (BH3D), extends dispersed structured light beyond the visible spectrum into the short-wavelength infrared (SWIR) band [1]. The system pairs a visible-spectrum silicon camera with an SWIR InGaAs camera in a stereo configuration, avoiding the multi-spectrograph designs that fundamental sensor constraints would otherwise require [2]. By modeling the image formation of broadband dispersed structured light, the authors estimate both hyperspectral reflectance and depth from a single capture [1]. Validation across diverse real-world scenes yielded a mean spectral angle mapper of 0.13 rad, a root mean square error of 0.03, and a mean depth error of 4.5 mm [1][2]. The broad spectral coverage exploits complementary physical cues: visible wavelengths capture surface appearance, while SWIR bands probe subsurface properties and material composition [2]. Spectroradiometry, the broader discipline underpinning hyperspectral imaging, measures how light energy is reflected, emitted, and scattered by substances to identify or differentiate materials [4]. Conventional geological applications of spectroradiometry operate within the visible-near infrared (400–750 nm) and SWIR (750–2500 nm) ranges, where absorption features tied to metallic oxides and clay minerals enable mineral identification [4]. The new BH3D method operates across a contiguous 450–1500 nm window, bridging the visible-near infrared and a portion of the SWIR band in a single system [2]. Tunable lasers, which can alter their output wavelength in a controlled manner, have long served spectroscopy and optical communications by providing narrow-linewidth sources across select bands [3]. The BH3D system instead uses dispersed structured light—a passive illumination strategy—to extract spectral information without requiring a tunable source [1]. The authors demonstrated practical capabilities including identifying metameric materials—objects that appear identical under one illuminant but differ spectrally—imaging through opaque layers, uncovering hidden features on banknotes, and revealing blood vessels [1][2]. The ability to image subsurface vasculature echoes the clinical utility of optical coherence tomography, which uses coherent near-infrared light and interferometry to obtain micrometer-level depth-resolved images of biological tissue [5]. While OCT relies on time-of-flight and amplitude detection of reflected light, BH3D reconstructs depth from structured-light triangulation combined with per-pixel reflectance spectra [1][5].