What Is Hyperspectral Imaging?

Hyperspectral imaging is an optical sensing technique that simultaneously captures spatial and spectral information across hundreds of contiguous, narrow wavelength bands, producing a three-dimensional data structure commonly called a hyperspectral cube. Each position in the two-dimensional spatial image plane contains a full spectrum, allowing the identification and mapping of materials based on their characteristic spectral reflectance or emission signatures rather than simply on brightness or broad-band color. This distinguishes hyperspectral imaging from conventional digital photography, which records only three broadband color channels, and from multispectral imaging, which typically captures fewer than twenty discrete bands.

The technique evolved from imaging spectrometry research at NASA's Jet Propulsion Laboratory during the 1980s, leading to the first operational airborne instrument, AVIRIS, in 1987. Spaceborne hyperspectral sensors followed, including NASA's Hyperion instrument on the Earth Observing-1 satellite (2000–2017) and more recent platforms such as the Italian Space Agency's PRISMA and the German EnMAP mission. Miniaturized sensor technology described by imec's hyperspectral imaging overview has since brought hyperspectral capability to handheld and drone-mounted platforms through on-chip filter integration.

Imaging Architectures

Hyperspectral cameras acquire spectral-spatial data using several scanning strategies, each involving different hardware designs and trade-offs. Whiskbroom scanners collect one spatial point at a time across the full spectral range, requiring a two-axis scan to build up an image. Pushbroom scanners, the most common design for airborne and spaceborne systems, image one spatial line across the full spectral range at each acquisition step, with platform motion providing the second spatial dimension. Snapshot or staring sensors capture the full spatial field of view in a single exposure, using spatial or spectral multiplexing schemes such as computed tomography imaging spectrometry or Fabry-Perot etalon arrays. Snapshot systems are essential for fast-moving scenes in industrial inspection or medical imaging, where target motion or frame rate requirements preclude scanning.

Spectral and Spatial Resolution Trade-offs

The spectral resolution of a hyperspectral system, measured in nanometers of bandwidth per channel, determines how finely the spectrum can be sampled. Typical hyperspectral systems achieve spectral resolutions of 5 to 20 nm in the visible and near-infrared, compared to bandwidths of 100 nm or more in conventional multispectral sensors. Narrower spectral channels contain less signal energy, creating a fundamental trade-off between spectral resolution and signal-to-noise ratio at a given integration time. Spatial resolution is simultaneously constrained by the detector array size, optics, and the data rate the system can sustain; airborne research systems may achieve sub-meter ground sampling distance while operational spaceborne systems typically range from 20 to 100 meters. The NASA Jet Propulsion Laboratory AVIRIS instrument description documents the design parameters of the reference airborne instrument, covering spectral coverage of 400 to 2500 nm across 224 channels with a 20 m ground sampling distance at typical survey altitude.

Calibration and Preprocessing

Before hyperspectral data can support material identification or quantitative analysis, raw digital numbers must be converted to physically meaningful units of radiance or reflectance. Radiometric calibration uses laboratory measurements of the detector's response to known light levels, correcting for dark current, non-uniform pixel sensitivity, and spectral smile (spatial variation in band center wavelength). Atmospheric correction, required for remote sensing applications, removes the contribution of atmospheric scattering and absorption to produce surface reflectance, using codes such as ATCOR or 6SV. Geometric correction registers the image to a geographic coordinate system using ground control points or direct georeferencing from GPS and inertial navigation data. The IEEE Xplore review on spatial enhancement of hyperspectral remote sensing imaging surveys preprocessing workflows and enhancement methods that improve downstream classification and unmixing accuracy.

Applications

Hyperspectral imaging has applications in a wide range of fields, including:

• Remote sensing for land cover mapping, vegetation monitoring, and mineral exploration
• Agriculture for crop health assessment, disease detection, and yield estimation
• Food safety inspection for contaminant detection and quality grading
• Medical and surgical imaging for tissue differentiation and cancer margin delineation
• Industrial quality control for detecting material defects and coating uniformity
• Art conservation and cultural heritage analysis for pigment identification