How can you tell if an avocado's gone bad just by looking at it? By examining it through a hyperspectral (HS) camera. These devices - also known as imaging spectrometers - see things the human eye cannot by scanning the world across multiple channels of light. Where humans see three wavelengths in the colour spectrum (red, green and blue), hyperspectral sensors can detect as many as 480.
Trained at an avocado - or any of a range of objects, from eyeballs to trees - they can make visible the item's traits, by using the reflective spectra created in the image's pixels.
Nasa developed the cameras 30 years ago for use in aerial imaging. "We're often in the business of trying to understand what things are made of without touching them," says Robert O Green, senior research scientist at Nasa's Jet Propulsion Lab. "Spectroscopy is a powerful analytical tool that enables remote measurement for scientific discovery and other applications."
These cameras once cost millions and weighed hundreds of kilograms - now, San Jose-based BaySpec sells 250g ones for $5,000 (£3,800). Sure, the quality isn't the same, as smaller devices tend to cover less spectral range, but as their price and size continues to fall, startups are sniffing an opportunity.
Carnegie Airborne Observatory, US
This California-based scientific conservation programme is mapping out the chemical properties of forests to help predict tree mortality and determine vulnerability to drought (WIRED 03.12). Two imaging spectrometers are fixed on an aeroplane to measure the water and nitrogen in each tree. "The hardware is stable enough to achieve global mapping," says project leader Greg Asner.
Gamaya's drone-mounted hyperspectral camera flies over farmland to map crops. The aim, according to co-founder Igor Ivanov, is to correlate spectral data with crop physiology to detect disease, weeds and nutrient deficiencies. So far, the team has mapped around 30,000 hectares of farmland in Brazil.
Quebec-based Optina Diagnostics is screening patients for early signs of Alzheimer's disease by taking hyperspectral images of their retinas. The team developed a metabolic hyperspectral retinal camera, which is being applied to identify beta-amyloids, a key biomarker of the disease. "The eye is a window into systemic illnesses," says David Lapointe, the company's CEO.
ImpactVision uses hyperspectral imaging to predict the shelf life of food. The startup, a product of Singularity University, is collaborating with the Fraunhofer Institute in Germany on a pilot to determine the pH and tenderness of beef. "There is a strong correlation between the pH value of red meat and its freshness," says CEO Abi Ramanan.
Based in Texas, Rebellion Photonics's spectral video cameras search for gases such as methane and hydrogen sulfide to prevent explosions on oil rigs, refineries and pipelines. When emissions exceed a certain threshold, the camera raises the alarm. "Our clients don't usually choose to [get the] alarm until a few thousand parts per million," says CEO Allison Lami Sawyer.
This article was originally published by WIRED UK
