Seeing the Invisible

How an Open-Source Imager Is Revolutionizing Science

The World Beyond Our Vision

Imagine looking at a forest and seeing not just trees, but their health status, water content, and species composition in vibrant, detailed colors.

This isn't a superpower—it's the revolutionary capability of hyperspectral imaging, a technology that has now been democratized through an ingenious open-source design.

Traditional Imaging

Captures only red, green, and blue wavelengths like human vision.

Hyperspectral Imaging

Records the full spectrum of light at each pixel, creating detailed "spectral fingerprints" 1 3 .

For decades, scientists have relied on hyperspectral imaging to capture data far beyond human vision. Until recently, this powerful technology came with a formidable price tag, often exceeding £20,000 for commercial systems 4 . That all changed with the development of the Hyperspectral Open-Source Imager (HOSI), a groundbreaking system that costs around £350—less than 2% of traditional equipment costs 4 9 .

98%

Cost Reduction

The Science of Seeing More

What Makes Hyperspectral Imaging Different?

Every material interacts with light in unique ways, absorbing some wavelengths and reflecting others to create what scientists call a "spectral signature." Where human eyes and conventional cameras see only broad color categories, hyperspectral imaging detects these precise signatures across hundreds of narrow wavelength bands 3 .

Hyperspectral imaging captures data across the full electromagnetic spectrum

This enables identification of materials based on their chemical composition rather than just their visible appearance. Think of it like this: where a standard camera might see a "green leaf," a hyperspectral imager can distinguish between a healthy leaf and a diseased one, detect water stress levels, and even identify specific nutrient deficiencies 1 6 .

The Open-Source Revolution

The high cost of commercial hyperspectral systems hasn't been the only barrier to widespread adoption. Traditional systems also tend to be bulky, complex to operate, and limited in their ability to handle high-contrast scenes 4 .

HOSI Performance Highlights
  • Minimum Light Sensitivity 0.001 cd.m⁻²
  • Dynamic Range >50,000:1
  • Cost ~£350

The HOSI system addresses these limitations through an ingenious redesign that combines off-the-shelf components with 3D-printed parts 4 . This makes it particularly valuable for researching artificial light at night (ALAN), a growing threat to global biodiversity 4 .

Inside a Groundbreaking Experiment: Mapping the Night

The Challenge of Artificial Light at Night

Artificial light at night has increased more than fourfold over the past 18 years, creating novel challenges for ecosystems worldwide 4 . Different animal species have varying visual sensitivities, and the impact of artificial light depends heavily on its spectral composition and spatial distribution.

Traditional satellite-based measurements miss crucial aspects of this light pollution, particularly horizontally propagated light and atmospheric effects like skyglow 4 . Understanding ALAN's true impact requires detailed spectral data collected at ground level across wide areas—exactly the challenge HOSI was designed to address.

Light Pollution Facts

Artificial light at night has increased more than fourfold over the past 18 years 4 .

HOSI Validation Results
Radiance Measurement Accuracy 98.01%
Reflectance Measurement Accuracy 97.5%

Methodology: How HOSI Captures the Night

The HOSI system operates on what's known as a "whisk broom" principle, building images point by point with a motorized gimbal that moves a compact Hamamatsu C12880MA micro-spectrometer across a scene 4 .

System Setup

The portable HOSI unit is positioned on a stable surface, with its operation controlled through a graphical user interface on a connected computer or smartphone 4 .

Scan Planning

The researcher defines the scan area and resolution, with the option to capture full panoramic images by combining multiple measurement points 4 .

Data Collection

The motorized gimbal moves the spectrometer through each predetermined point in the scene. At each position, the system captures the complete spectral data from 320–880 nm with a spectral resolution of approximately 9 nm (FWHM) 4 9 .

Radiance Calibration

The raw data is converted to absolute radiance values using calibration factors determined through the system's calibration process 4 .

Image Construction

The individual spectral measurements are compiled into a hyperspectral "data cube"—a three-dimensional representation with two spatial dimensions and one spectral dimension 4 .

Results and Analysis: Revealing Hidden Patterns

In validation tests, HOSI demonstrated impressive accuracy, with mean absolute errors of just 1.99% for radiance measurements and 2.5% for reflectance compared to professional-grade instruments 4 .

Parameter Specification Significance
Cost ~£350 Makes hyperspectral imaging accessible to researchers, educators, and citizen scientists
Spectral Range 320–880 nm Covers ultraviolet, human-visible, and near-infrared wavelengths
Spectral Resolution ~9 nm (FWHM) Sufficient to distinguish fine spectral features of different light sources and materials
Minimum Light Sensitivity 0.001 cd.m⁻² Enables measurement in low-light conditions like moonlit environments
Dynamic Range >50,000:1 Allows capture of scenes with both very bright and very dark areas
Spatial Resolution ~2 cycles per degree Detailed enough for environmental mapping and light distribution analysis

Applications Across Science and Society

The implications of affordable, accessible hyperspectral imaging extend far beyond ALAN research. The same technology that maps light pollution can also transform fields from agriculture to medicine.

Environmental and Ecological Monitoring

Hyperspectral imaging has proven invaluable for forest classification, soil analysis, and water quality assessment. One study noted that hyperspectral satellites improved forest classification accuracy by up to 50% compared to conventional methods 1 .

The technology can also detect marine plastic waste with 70-80% accuracy and map soil organic matter content, providing crucial data for combating pollution and managing natural resources 1 .

Agriculture and Food Safety

In agriculture, hyperspectral imaging enables early detection of crop diseases before visible symptoms appear. The HSI-TransUNet model achieved 98.09% accuracy in detecting crop diseases and 86.05% in classification, potentially revolutionizing precision farming 1 .

For food quality assessment, the technology has predicted egg freshness with an R² of 0.91 and achieved perfect 100% accuracy in pine nut quality classification 1 .

Medical Diagnostics

Perhaps one of the most impactful applications is in healthcare, where hyperspectral imaging can differentiate between healthy and cancerous tissues with high sensitivity and specificity—87% and 88% for skin cancer, and 86% and 95% for colorectal cancer detection 1 .

The non-invasive, label-free nature of the technology makes it particularly suitable for clinical use and real-time decision support during surgeries 1 .

Field Application Reported Effectiveness
Healthcare Cancer detection (skin, colorectal) 87% sensitivity, 88% specificity (skin); 86% sensitivity, 95% specificity (colorectal)
Agriculture Crop disease detection 98.09% accuracy in detection, 86.05% in classification
Food Safety Egg freshness prediction R² = 0.91
Environmental Monitoring Marine plastic detection 70-80% accuracy
Pharmaceutical Counterfeit drug detection Accurate identification of fake anti-malarial tablets
Waste Management Material identification for recycling Robust characterization of complex multi-material objects

The Scientist's Toolkit: HOSI Components and Their Functions

Building a HOSI system requires both hardware components and software tools, all designed to be accessible and modifiable by the research community.

Component Function Specifics in HOSI
Micro-spectrometer Captures detailed spectral information at each point Hamamatsu C12880MA spectrometer
Motorized Gimbal Precisely positions the spectrometer for spatial scanning Custom system with stepper motors for 2-axis control
Control Electronics Coordinates system operation and data collection Arduino-based controller with custom firmware
Structural Components Houses and supports all system elements 3D-printed parts designed for off-the-shelf components
Calibration Materials Converts raw data to absolute radiance values Spectralon standards for reflectance calibration
Software Interface Controls operation and processes data Graphical user interface compatible with computers and smartphones

The complete design, including all code, calibration data, and 3D printing files, is available through a GitHub repository and Zenodo archive, ensuring that researchers worldwide can not only use but also modify and improve the system 4 .

A Brighter, More Accessible Future

The development of the Hyperspectral Open-Source Imager represents more than just a technical achievement—it demonstrates how open-source principles can democratize powerful technologies that were previously accessible only to well-funded institutions.

By reducing the cost of hyperspectral imaging from tens of thousands to hundreds of pounds, HOSI opens doors for researchers, educators, and even citizen scientists to explore the spectral world in ways previously unimaginable.

As artificial intelligence and machine learning continue to advance, the combination of these technologies with accessible hyperspectral imaging promises even greater breakthroughs. Researchers are already developing AI-driven analysis techniques that can automatically interpret complex hyperspectral data, identifying patterns and making predictions that would challenge human analysts 1 2 .

Democratizing Science

Open-source tools like HOSI make advanced research accessible to all

The invisible world of spectral information surrounds us every moment, filled with insights about our health, our food, and our environment. Thanks to open-source innovations like HOSI, we're all gaining the ability to see this hidden dimension—and what we're discovering promises to transform our understanding of the world around us.

References