The Living Optics Hyperspectral Camera allows developers to capture spectral information at video frame rates, opening up new application areas and significantly improving precision in computer vision applications such as detection, identification, segmentation and object tracking.
RGB sensors used in standard cameras only have three values per pixel. Hyperspectral imaging stretches what it is possible to see by detecting many more bands of light. In this case, it records 96 data points per pixel and expands beyond the visible spectrum, providing information that leads to a number of other properties such as the material it is made of.
A study of the underlying optics of Living Optic’s technology led to the novel L-shaped form which is easily recognisable and scales to different sizes as well as being a flexible enough form to cope with different approaches to the lens component allowing evolution over time as the optics were refined
Living Optics approached Pentagram with a very specific need—to take the core technology that had been demonstrated in a functional prototype and develop a design language that could be applied to a first generation product/development kit and then later to manufactured units.
Anodised aluminium was chosen as a key material as it allows the design to be CNC machined, developed and refined accross different manufacturing batches. It references both camera and computing archetypes.
Living Optics, headquartered in Oxford, UK is a pioneer in the development of cost-effective, high frame-rate hyperspectral imaging systems, which provide more than 10 times the data captured by traditional RGB cameras.
Whilst humans feel we can see a vast spectrum of colours, in reality, our eyes and most cameras contain three types of colour receptors sensitive to red, green and blue light. This means that every colour we see is the result of our brains balancing the relative RGB light detected by the receptors in the eyes. Hyperspectral cameras allow the capture of colour in between and outside of that spectrum, stretching what it is possible to see by detecting many more bands of light, offering 30 or more data points per pixel and expanding beyond the visible spectrum. By generating more spectral information we learn much more about the subjects of an image. A forthcoming feature will allow the hyperspectral imaging to provide information about temperature – this has many potential uses, such as determining an object’s material composition, or measuring a patient’s concentration of red blood cells.
The output of hyperspectral cameras is not visible to the human eye as an image, but software can be used to interpret the data and help detect very specific and precise colours within the spectrum, showing things like the health or ripeness of a plant or crop that is normally invisible to the human eye. The technology allows for significantly improved precision in computer vision applications such as detection, identification, segmentation, and object tracking as well as non-invasive diagnostics in healthcare.
Hyperspectral cameras have been available for a number of years however they are typically very bulky, high cost and have extremely low frame rates. Living Optics’ patented technology uses a combination of hyperspectral reconstruction neural networks, coded aperture designs, and optical architectures to deliver hyperspectral performance at video frame rates in a small form-factor at significantly lower cost than the current market. The technology is optimised for mass production and makes use of widely available and cost-efficient CMOS sensors, eliminating the costly need to customise a sensor.
Living Optics approached Pentagram with a very specific need – to take the core technology that had been demonstrated in a functional prototype and develop a design language that could be applied to a first generation product / development kit and then later to manufactured units. The process was highly collaborative with the Pentagram team working directly with Living Optics founder and CEO Robin Wang and team on a weekly basis to develop, refine and test the designs.
A study of the underlying optics of Living Optics’ technology led to the novel L-shaped form which is easily recognisable and scales to different sizes as well as being a flexible enough form to cope with different approaches to the lens component allowing evolution over time as the optics were refined.
Anodised aluminium was chosen as a key material as it allows the design to be CNC machined, developed and refined across different manufacturing batches. It references both camera and computing archetypes. The aim was to respect some of the codes and signals of optics and precision that are present in most cameras but add a layer of purity and simplicity that reflected the software-driven experience.
The camera is the heart of the Living Optics Development Kit alongside software tools for camera control, data exploration and analysis to allow programmers to explore hyperspectral imaging for computer vision applications. The camera itself is a hero device but the design approach acknowledged that it exists as part of an architecture including edge compute, field accessories, protective case, tripod, lighting and other accessories.
A key aspect of the project was the development of the camera to launch with the Living Optics Development Kit camera as well as roadmap development for future products with smaller form factors and different materiality, all the time retaining the core Living Optics design language.
Office
- London
Partner
Project team
- Shing Lo
- Chifen Cheng
- Yemima Lorberbaum
- Amelia Kociolkowska
- Christie Sung