PARC Scientists Develop Tiny, Low-Cost Hyperspectral Imaging Cameras

Estimated reading time: 2 minutes

Hyperspectral imaging is currently used in a variety of industries, including remote sensing, agriculture, biomedical imaging, geosciences, machine vision, surveillance and security, and defense. All objects leave fingerprints in the electromagnetic spectrum known as spectral signatures that offer information about a scene, such as material composition, whether a target is present, or the state of objects in the scene. Despite the compelling ability of hyperspectral cameras to record the spectral signatures within a scene, they have found a limited market due to their lavish price tags, with costs ranging from $50K at the low end to well over $1M. As a result, the data analysis needed to understand hyperspectral images requires trained experts and has not been automated or standardized.

Researchers at PARC are democratizing hyperspectral imaging with a fundamentally new approach that dramatically reduces the associated cost and size barriers. As a result, we will soon see hyperspectral cameras on mobile phones and other handheld devices. By targeting such a low price point, industries that use hyperspectral imaging will be able to greatly expand both the operational and research use of these cameras because they’ll be able to deploy many more devices across their applications and use cases.

The software flexibility and wide wavelength range afforded by the PARC technology means that new applications can be fully developed in software without any of the hardware changes required with existing lower-cost technologies. PARC’s work is enabling a general-purpose hyperspectral imaging platform that can be leveraged by mobile app developers, which is paving the way to make this extremely advanced technology more affordable to average consumers.

Imagine if you could use your cell phone at the grocery store to determine if meat will be tender or if fish was previously frozen, or to find the ripest piece of fruit. You could take a photo of a suspicious mole for an immediate, automated diagnosis to be uploaded to your doctor. At home, imagine taking images of the flowers in your garden, and an automated app tells you how much to water or fertilize. Stain your clothing at dinner? Take a picture to receive an instant remedy. 

All these applications are enabled by the addition of a liquid crystal layer about as thick as one human hair to an existing image sensor, and PARC is creating it to cost less than your daily coffee. 

Just as a dog has many more scent receptors than a human and therefore has an ability to distinguish scents that humans cannot (humans rely on dogs as advanced sensors for all sorts of tasks!), a hyperspectral camera has many more “wavelength” detectors than the human eye. Combined with the number-crunching ability available in the cloud to make sense of all the wavelength data, a hyperspectral camera in everyone’s pocket has tremendous value.