FEATURE: HYPERSPECTRAL


“Our goal is to make it super easy for anyone to extract information from that kind of data”


emergence of combined high spatial and spectral resolutions with 5-metre GSD and 250 bands from our partner, Pixxel. I believe everyone throughout the Earth observation industry is excited to dip into this new data.” Part of this improvement comes from


advances in miniaturised hyperspectral sensors for small satellites, which enhance accessibility and affordability, explains Aakash Parekh, CCO at Pixxel. “This will invariably promote the


democratisation of hyperspectral imaging technologies across all industries. As technology advances, so does the sensors’ spectral and spatial resolution. This enables even more insights into never-before-seen characteristics of the Earth’s surface.” Pixxel and EUSI have partnered to


capture up to 250 bands at 5m resolution with daily revisit over Europe, setting a new benchmark for the region. But it’s not just the hardware. Much


of the current progress in the field also comes on the software side.


www.imveurope.com | @imveurope


Processing in space, AI on Earth Hyperspectral imaging produces a lot of data. Cameras record tens or hundreds of different spectral bands for each pixel. This makes the volume and computing requirements for storing, processing, and analysing this data substantial, says Goodman. Meeting this challenge requires advances in many aspects of the hyperspectral data processing chain, from image acquisition and downlink to data analysis and product delivery. Everything, from getting the data down to Earth, to its processing and interpretation, presents challenges. Processing data onboard is one


approach, recently discussed at Space Tech Expo Europe in Bremen. Some technical solutions for preliminary onboard data processing exist and have been discussed in the academic literature. The industry is also taking steps in this direction.


HySpeed and Metaspectral are


preparing a payload for the International Space Station (ISS) that will demonstrate real-time streaming capabilities and analysis of ultra-high-resolution imagery from space to ground. The payload is called Onboard Programmable Technology for Image Compression and Analysis (OPTICA) and is scheduled for launch in


2024. OPTICA will include hyperspectral sensors coupled with real-time data analysis hardware. Previously, HySpeed had deployed hyperspectral image processing software, also aboard the ISS. This is “the first commercial off- the-shelf edge computing system with AI capabilities built for in-space data processing”, says Goodman, and can be used in ground observation as well as lunar or Martian studies. Cloud-based data processing and


analysis is another option, as is data processing and interpretation here on Earth. In the latter, AI is also bringing new impetus. “Our core product is really on the


analysis side,” says Francis Doumet, founder and CEO at Metaspectral. “We’re using AI to basically extract information that would otherwise go undetected or that would be hidden without AI techniques.” Another challenge presented by large


volumes of data is the difficulty of analysing it. Historically, says Doumet, you would have needed a PhD to make sense of the data, and it would take days to process a dataset. In this regard, he believes AI can truly be a game-changer. “We’re getting the software ready to


be able to analyse data from any of the hyperspectral constellations being put in orbit right now,” he says. “Our goal is to make it super easy for anyone to extract information from that kind of data. The benefit of our software is it allows anybody, with or without a technical background, to start training AI models on the data. And you can train a model in as little as 10 minutes.” This is even more important as


applications of hyperspectral imagery expand and become available to more and more consumers.


DECEMBER 2023/JANUARY 2024 IMAGING AND MACHINE VISION EUROPE 13


Shutterstock/Andrei Armiagov


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32