You might not realize it, but the range of light and color visible to the human eye is actually quite limited when compared to all the things we can’t see. There’s a huge chunk of the light spectrum that’s completely invisible to our highly-evolved eyeballs — but luckily we’ve got technology that can fill the gap. HyperCam, for example, is a hyperspectral camera that captures images in the full range of light from visible to near-infrared in order to analyze surfaces, textures, and even human skin. Hyperspectral imaging technology isn’t exactly new, but HyperCam is the first step towards making the tech cost-effective and scalable.
Satellite imaging, infrastructure and energy analysis systems and food safety inspections all use hyperspectral imaging, but the technology is still prohibitively expensive for use outside industrial purposes. Researchers at the University of Washington (with help from Microsoft’s research branch) set out to make an affordable hyperspectral camera that could theoretically be used in smartphones and other consumer gadgets. The researchers admit the technology isn’t quite ready for consumer applications yet, but demonstrations show that it’s also not far off. The hardware solution prototyped by the team could cost as little as $800, and has been envisioned as an eventual smartphone add-on at no more than $50.
In hyperlapse video footage, HyperCam has been used to show beneath the skin of ripening fruit and vegetables. The visuals demonstrate the information captured by HyperCam that is invisible to the human eye, and could mean never cutting into a not-quite ripe avocado again. Projections for consumer use of HyperCam technology would mean that grocery store shoppers would be able to scan displays of fresh produce using their smartphones, detecting imperfections or even dangerous inclusions in foods before carting them home from the store.
The implications for medical innovation are perhaps more compelling, from a diagnostic health perspective in addition to the technology behind those possibilities. HyperCam captures 17 different wavelengths of light, generating a unique image at each wavelength. Researchers knew immediately that sorting through seventeen images for each instance would not be a scalable solution, so they developed software to analyze each set of images for the most information-rich results. When hooked up to UW’s custom built software, HyperCam is able to identify and select preset wavelengths that are particularly telling for a specific image sample.
In demonstrations with a human hand, the software selected the wavelengths whose images contained the most information, in this case visuals of veins not detectable by the human eye but suddenly clear as day using HyperCam imaging. As other technologies progress separate from HyperCam, it’s easy to see how the imaging tech could be used in collaboration with smart algorithms, neural networks, and artificial intelligence programs. Keeping costs down on the imaging technology opens HyperCam to a range of smart tech-enabled possibilities that will only enhance its imaging advantage.
