With its low cost, compact size, and simplicity, a new hyperspectral imager from SRI improves situational awareness in adverse conditions.
Human eyes are impressive but limited instruments: Our cone cells (the eye’s color receptors) are most sensitive to light in three narrow bandwidths corresponding to the colors red, green, and blue. Ordinary cameras mimic the human eye, capturing limited information within those three wavelength bands.
Hyperspectral imaging provides a way around this physiological limitation. Unlike ordinary cameras, hyperspectral imagers are designed to sample numerous wavelengths across a band that generally includes — and often extends beyond — the entire visible spectrum. By translating this additional data into information that users can perceive, hyperspectral cameras are being used to identify camouflaged military vehicles, monitor crop health, diagnose medical conditions, and even help museums detect art forgeries.
However, the high cost and bulky nature of hyperspectral imagers has limited their accessibility, practicality, and adoption.
To better reap the benefits of the technology, SRI researchers have developed a cost-effective new kind of hyperspectral imager with many applications across the civil and defense markets.
How hyperspectral imaging works
Hyperspectral imagers are distinguished from other imaging technologies by their approach to pixels.
Whereas a traditional image pixel presents a single color in a defined space, hyperspectral pixels are more information-rich. “The key concept in hyperspectral imaging is that each pixel in an image contains not only spatial information, like a regular camera, but also spectral information across a range of wavelengths,” explained Joerg Martini, associate director and area manager of Autonomous Systems and Life Sciences at SRI.
The resulting, processed image cube gives a fuller, more nuanced picture of the observed world. Users, as well as image-recognition software, can view these enhanced images and readily pick out anomalies, such as the discrepancy in infrared light reflectivity between the camouflage netting over an enemy military vehicle and the paint on the vehicle itself.
Designing a more practical hyperspectral imager
The SRI hyperspectral imager is conveniently smaller and less expensive compared to other available devices, while still providing sharp image resolution and capturing information at wavelengths across the visible spectrum and up to about 1050 nanometers in the infrared range. The imager can be easily deployed in applications where size, weight, and power availability are constrained, for instance on drones.
“Our system’s simplicity, compactness, low cost, and performance make it a promising candidate for a wide variety of applications where traditional, larger, and more expensive hyperspectral imaging systems might not be practical,” said Martini.
“Our imager does efficient data collection without the need for mechanical components or moving parts, making the system robust to vibrations.” — Joerg Martini
On a technical level, SRI’s new hyperspectral imager relies on a liquid crystal variable retarder, a technology that has rarely been employed in hyperspectral imagers. Liquid crystals are substances that have some degree of order, like crystals, but can still flow like liquids and respond to external control stimuli. Liquid crystals can dynamically change the polarization of light, meaning the plane in which the electric field of light oscillates. When sending light of a certain polarization through a polarizer, light can be blocked or transmitted depending on orientation, just like when the screen of a smartphone becomes black when looked at through polarizing sunglasses.
Taking a cue from old-school pocket calculators, which use a liquid crystal display to switch digits on the display between transparent and black with electrical signals, the scientists at SRI employed the same basic method (though in a slightly more sophisticated way) to block and transmit light of certain controllable wavelengths. The result is a pattern of light that can be mathematically parsed by software to harvest spectral information. Although this processing task is complex, it takes only two seconds to complete, quickly producing a set of images containing potentially critical spectral as well as spatial information.
“The image sequence lets you see things you otherwise couldn’t,” said Martini.
Because liquid crystal technology is both mature and inexpensive, it turns out to be ideal for creating a hyperspectral imager that is small, light, and cost-effective.
As an additional engineered benefit, because the SRI imager (unlike many high spectral resolution imagers) does not have any moving parts, it is better able to withstand the rigors of many plausible working environments. “Our imager does efficient data collection without the need for mechanical components or moving parts, making the system robust to vibrations,” said Martini.
Why inexpensive, practical hyperspectral imaging matters
The SRI hyperspectral imager lends itself to many potential use cases. In particular, defense looms large.
Different militaries and armed groups use distinct textiles and paints. Based on these distinctions, hyperspectral imagers deployed on reconnaissance platforms can determine the national identity of deployed military vehicles, aircraft, and even satellites.
“The military has figured out that, both from space and from air, they’re able to identify friend or foe using hyperspectral cameras,” said Dan Williams, a business development executive at SRI and a former member of the U.S. Army. “When it comes to foreign aircraft and drones, with our imager you’re able to make some level of differentiation of whether you should let that thing keep flying or shoot it down.”
Hyperspectral cameras provide a powerful means of defeating camouflage, which is typically tailored to conceal visible patterns. “Hyperspectral cameras give you a high level of accuracy in identifying anomalies — things that aren’t supposed to naturally be there in an environment,” said Williams. Thanks to its manageable size and cost, SRI’s imaging technology can be distributed much more widely within the armed forces, allowing units to take advantage of a new level of environmental awareness.
These sensing capabilities extend into numerous commercial use cases, including industrial inspection, counterfeit surveillance, and agricultural yield management. For the latter, Martini explained, “A lot of plant physiology manifests itself in the color of the plant, fruit, or seeds, and there are many subtle physiological features that the human eye doesn’t pick up on due to its limited spectral sampling range.”
Another promising application for the SRI hyperspectral imager is fluorescence microscopy, which involves shining light on fluorescent markers in lab samples to reveal specific elements of interest through highly detailed visualizations, often used in biology and medical diagnostics. Yet another use case is self-driving cars, where hyperspectral imagers could equip autonomous vehicles to more precisely discover and classify objects in their surroundings.
“We’ve been developing this imager to occupy a particular area in the marketplace,” said Martini. “Where high spatial resolution, small size, low cost, simplicity, and robustness are critical features.”
To learn more about SRI’s hyperspectral imager or other technologies we are developing for government and commercial clients, contact us.
