Subscribe via E-mail

Your email:

Follow Us!

Headwall Photonics Blog

Current Articles | RSS Feed RSS Feed

UAVs and Hyperspectral Imaging Unite

 

One of the things we’re seeing at Headwall is the proliferation of airborne applications. Multispectral suffers a bit with respect to hyperspectral (a handful of bands versus hundreds), which is why hyperspectral is winning the day.

UAV choicesOne reason is instrument affordability. Multi-million-dollar hyperspectral sensor programs might have flown (literally and figuratively) in the military world, but not in precision agriculture or with universities. Budgets are smaller, and that money has to be spread among not only the sensor but the UAV and everything in between. This is where small, entrepreneurial companies like Headwall shine, because everything in between can mean LiDAR, GPS/IMU technology, application software, data processing, and so much more. We understand hyperspectral imaging better than anyone, and our focus has always been to better that technology while driving costs lower. This is the essence of commercial-off-the-shelf (COTS), where highly specialized military instrumentation finds a home all across industry and academia. With respect to Headwall, COTS implementation means smaller, lighter and more affordable sensors that are easier to use yet just as optically precise as their multimillion-dollar military counterparts.

Second, you cannot go a day without seeing stories about UAVs. Fixed-wing designs like those from AGX and PrecisionHawk are crowding the skies along with multi-rotor helicopters like Infinite JIB and AIBOTIX. These are much more than hobbyist playthings and are perfect for scientific reasearch duties. They have excellent range and payload-carrying characteristics, and they are stable aloft.  From mineral exploration and agriculture to petroleum and pollution control, UAVs are everywhere it seems. And everyone takes notice when household names like Facebook, Google and Amazon decide that the UAV is going to be instrumental to their future success. Much of this might sound fanciful and far-off, but it is happening now. Court challenges are being won, and while care needs to be taken on how regulations are drafted and enforced, no one doubts that the UAV is not only here to stay but will become commonplace.

Obviously, UAVs simply take up airspace unless they are doing good work. And largely, we seem to hear about bad things happening when mention of UAVs (and drones) is made. But stop and consider for a moment how a famine-stricken area can be made crop-fertile thanks to hyperspectral data that a UAV-mounted sensor can collect. A scientist will know about disease conditions with enough time to prevent damage by skimming the treetops and looking for anomalies that become ‘visible’ through hyperspectral imaging. A farmer will know where to plant and harvest…and where not to. Crop stress will be seen long before it becomes a worry, and the amount of wholesome and nourishing food planted in areas once thought impossible will blossom. In short, small and light UAVs are affordable for the people who need to use them. They can be flown in areas that vehicles and humans cannot yet reach, providing a window of research never available to scientists before.

As we see the proliferation of UAVs capable of carrying sensor payloads, it is important to understand how everything goes together. Here, Headwall is taking a leading role. Many mistakenly believe that slapping a sensor onto an octo-copter is all they need to do. But making sure everything works the way it should aboard a flying, unmanned vehicle is another challenge altogether. How much ground do you need to cover, and do you have enough battery power to do it? How much hyperspectral data do you need to collect, and do you have the computing and storage horsepower to make that happen? What are you looking for, and what spectral ranges are those things in? How do you ortho-rectify the data during post-processing? And how do you use the science of ground-truth as it relates to airborne hyperspectral imaging? This last consideration is hugely important, because the collaboration of airborne hyperspectral and ground-truth delivers the best possible accumulation of data. Headwall and ASD have even authored a 12-page whitepaper on the relationship between airborne hyperspectral data and ground-truth techniques.

Headwall's Field-of-View Calculator

 

When it comes to hyperspectral imaging, it isn’t always about the hardware. Before users even get to the stage of specifying a sensor instrument, they need to ask a few questions:

  • What do I want to look at?
  • How am I deploying the sensor?
  • What is the spectral range of what I’m looking at?
  • How far from the object will I be?

The answers to these questions will lead to an informed decision about the kind of sensor that’s best, the kind of lens it will need, and how small and light the sensor needs to be.  At Headwall, we’re helping customers sort through these questions and considerations every day. We make on-line tools available that make instrument specification easy. With the answers to a few simple questions, the overall application-specific design of a hyperspectral instrument is well within reach. This means quicker time-to-deploy for customers who have challenging scientific questions that need answers.

One of Headwall’s newest tools is the Field-of-View (FOV) calculator. This tool collects a few important user-defined parameters to arrive at several what-if scenarios. The first parameter is distance from lens to object. In an airborne application, the distance would likely be measured in meters. For lab-based or in-line deployment, it might only be centimeters.  The second parameter is the wavelength, which can be UV-VIS (380-825nm) all the way up to SWIR (950-2500 nm). Knowing the spectral signature of the item of interest will point you in the correct direction.

FOV resized 600

The calculator will take this information and combine it with choice of sensor and lens to arrive at useful data for the customer. In this case, we see that for the parameters and options chosen we are given the number of spatial and spectral channels (1004 and 335 respectively). We’re also given the linear and angular FOV, the instantaneous FOV, and the spectral resolution. In an airborne application, the linear FOV can be thought of as the flight swath. The wider the better, because the aircraft or UAV will be able to collect full hyperspectral information with fewer passes over the ground.

Output

Spectral libraries are common starting points for defining where to look along the spectral range. The spectral signature for everything from plants and crops to minerals and petroleum is known or catalogued.  While everything has its own signature, the real strength of hyperspectral imaging is to discriminate and classify. So while the sensor can actually ‘see’ everything, it is tuned to look for things that may resonate at 900 nm or 1900 nm for example. A disease condition on a fruit tree may be impossible to detect by any visible means, but it will resonate quite clearly when seen with a hyperspectral sensor.

Customers come to Headwall regularly with certain ‘needs.’ A crop scientist may want to analyze the soil from an airborne UAV. Another may want to adopt hyperspectral imaging along a high-speed food processing line to see and remove foreign matter. A third may be a museum preservationist interested in understanding the artwork and artifacts under their care. But in all cases, the first question is: What do you want to see?

 

History Made, History Seen with Hyperspectral Imaging

 

As the market for hyperspectral sensing technology moves forward and advances, Headwall’s Application Engineering team has been able to gather a rare view into the past through the hyperspectral scanning of some of the most important historical artifacts and papers in the United States. For the first time ever, hyperspectral VNIR and SWIR imaging was conducted on key historical documents from the US Civil War period.

The Gettysburg AddressBy working collaboratively with the researchers in the Cornell University Division of Rare and Manuscript Collections and the Cornell Johnson Museum of Art, Janette Wilson and Kwok Wong of Headwall’s Application Engineering team spent a few days conducting VNIR and SWIR hyperspectral scans of some of the most important artifacts held by Cornell University. Of particular interest was the hyperspectral scanning of the University’s collection of original Lincoln documents signed by president Abraham Lincoln during his presidency. This collection included the Gettysburg Address (seen at left), the Emancipation Proclamation, and the 13th Amendment to the Constitution.

The scanning of documents and artifacts with hyperspectral imagers is particularly well suited for the purposes of both 1) research and 2) for establishing a baseline of spectral/spatial information for monitoring change in the artifacts to better preserve objects of cultural heritage.

For a couple main reasons, hyperspectral imaging is particularly appealing to collection-care experts. First, and probably most important, is that the technology is non-destructive. The instruments don't interface with the documents and the lighting is called 'cold illumination.' That is, there is no risk of themal damage to the items under inspection. Second, previously unseen features immediately 'come to light' when viewed hyperspectrally. Note the image below, which represents a stamp on the Gettysburg Address that cannot be seen visibly but can when looked at within the VNIR and SWIR spectral range. Collection-care experts are fascinated by unseen features, which can be used to build the body of knowledge with respect to documents or artifacts.

Unseen Features

Hyperspectral Sensors for UAV Applications

 

The scientific research community is beginning to understand and embrace hyperspectral imaging as a useful tool for a few primary reasons. First, sensors are more affordable than ever. Originally conceived as multi-million-dollar ISR platforms for defense applications, hyperspectral imagers have been successfully ‘commercialized’ over the past few years. Scientists typically embracing RGB or multispectral technology before can now acquire hyperspectral sensors at affordable price points.

Hyperspectral sensors of the ‘pushbroom’ type produced by Headwall require motion to occur. That is, either the sensor flies above the field of view, or the field of view moves beneath the sensor. For UAV applications, Headwall’s small and lightweight Micro-Hyperspec is the platform of choice. Available in the VNIR (380-1000nm), NIR (900-1700nm), and SWIR (950-2500nm) spectral ranges, the sensor is truly ‘SWaP-friendly.’

Spectral range is often where the decision-making starts. The chemical fingerprint—or spectral signature—of anything within the field of view will lead the user in one direction or another. For example, a certain disease condition on a tree canopy may become ‘visible’ within the SWIR spectral range (950-2500nm). Similarly, a certain mineral deposit may become ‘visible’ in the VNIR range (380-1000nm). One approach to ensuring the spectral ‘fidelity’ of images collected by the sensor makes use of ‘diffractive optics’ comprising aberration-corrected holographic gratings. This ‘Aberration-corrected concentric’ design is shown below.

concentric imager

There are several advantages to this ‘reflective’ approach. First, the design is simple, temperature insensitive, and uses no moving parts. This assures robustness and reliability in airborne situations. Second, diffraction gratings can be made very small so that the instruments themselves can be small and light; in other words, capable of fitting the new class of lightweight, hand-launched UAVs. Third, the design optimizes technical characteristics that are most important: low distortion for high spatial and spectral resolution; high throughput for high signal-to-noise; and a tall slit for a wide field-of-view. Because the design is an all-reflective one, chromatic dispersion is eliminated and excellent focus is assured across the entire spectral range.

Many within the environmental research community and across ‘precision agriculture’ prefer to use UAVs as their primary airborne platform. They are more affordable than fixed-wing aircraft and easy to launch. But as UAVs get smaller and lighter, so must the payloads they carry. And integrating the sensor into the airframe along with other necessities such as LiDAR, power management/data collection hardware, and cabling can be a daunting task (Figure 3). Orthorectification of the collected data is another key requirement, which is the means by which the hyperspectral data cube is ‘managed’ into useful information that has been ‘corrected’ for any airborne anomalies. In other words, the collected hyperspectral data needs to be ‘true’ to what’s actually within the field of view.

 Micro Hyperspec

Acquiring a UAV and a hyperspectral sensor won’t assure compatible performance, and a high level of ‘integration work’ is needed. The UAV community and the hyperspectral sensor community are both challenged with pulling everything together. Recognizing this, Headwall Photonics is taking an industry-leading position as a supplier of fully integrated airborne solutions comprising the UAV, the sensor, the power and data management solution, cabling, and application software. The result is that users are flying sooner and collecting better hyperspectral data than ever before.

Type of UAV is very often one of the first decisions a scientist will need to make. Fixed-wing and multi-rotor are the two general categories, with numerous styles and designs within each. In-flight stability and flight-time duration are both paramount concerns, and this is where payload restrictions will often point toward one or the other. Multi-rotor UAVs launch and land vertically, so this type will be favored in situations where space is tight. Conversely, a fixed-wing UAV requires suitable space to launch and land but can provide longer flight duration and carry a heavier payload. The wide field-of-view characteristic of the concentric imager allows a UAV to ‘see’ more ground along its flight path.

Integrated airborne package

Two other key areas managed through Headwall’s integrative process are data management and application software. While a separate subsystem is used to control the sensor operation and store the hyperspectral data, the direction is clearly toward on-board integration of these capabilities. Flash storage and solid-state drives will soon make it possible for the sensor to ‘contain’ all the related functionality that now needs to be contained in a separate module. This will clearly lighten the overall payload, reduce battery consumption, and boost airborne flight time.

Headwall’s Hyperspec III software represents a complete, modularized approach to the management of hyperspectral data. Orthorectification is one such module within the software suite that removes the unwanted effects airborne behavior. The resultant orthorectified images have a constant scale wherein features are represented in their 'true' positions. This allows for the accurate direct measurement of distances, angles, and areas. Other aspects of the software suite can be used to control GPS/IMU devices, control multiple sensors simultaneously, and save polygons (A Google-map-enabled tool that allows the user to define geographic coordinates).

 

 

Bone-Dating Using Hyperspectral Imaging!

 

Headwall recently completed some fascinating demonstration work on behalf of the Conservation Manager and several colleagues at London's Natural History Museum.

One of the hallmarks of hyperspectral imaging is its ability to non-destructively and non-invasively collect an invaluable amount of spatial and spectral data from any sort of reflected matter within the field of view. In commerce and environmental studies, hyperspectral imaging is a valuable and well-known tool that can ‘see’ the unseen.

composite bonesForensics is another exciting area of research. Take 300,000-year-old Neanderthal human bones, for example. Or a 300-year-old snake skin. Or a 400-year-old book of poems. Here you see two bones within the field of view of Headwall's VNIR starter kit. The smaller one is 'only' 200 years old; the larger is 300,000 years old. But the beauty of hyperspectral sensing is that it can classify and compare specimens like these with a tremendous amount of precision, yielding a level of scientific analysis that museums and 'collection-care' experts crave. The demonstration that Headwall performed was an exciting opportunity to show off not only the capabilities of the sensor, but also the capabilities of our new Hyperspec III software. The Conservation Manager was extremely excited with the results of the demonstration. He even remarked that his museum would like to embrace and move forward with the opportunity to be a 'Centre of Excellence for Hyperspectral Imaging,' with Headwall as its sponsor.

Spectral ‘fingerprints’ contain a tremendous amount of useful data, and hyperspectral instruments can see these fingerprints and then extract meaningful data regarding the chemical composition of anything within the field of view. More helpful still, these instruments work in tandem with known spectral libraries that allow a very high degree of selectivity and discrimination. If you know the spectral fingerprint associated with a particular chemical, you can reference it against the hyperspectral data cube coming from the sensor. That fingerprint, once found, very often will be a ‘predictor’ of something else. Disease conditions in crop trees, for example, or the presence of certain inks or pigments on a document or artifact. That’s why precision agriculture and document verification are two other common deployment areas for hyperspectral imaging.

Headwall Remote Sensing Capabilities Seen “Down Under”

 

melbourneThis past week, Headwall remote sensing team finished a productive week Down Under at the International Geoscience and Remote Sensing Symposium (IGARSS) in Melbourne, Australia.  The conference, organized by the IEEE, comprises a ‘Who’s Who’ across the global remote sensing community. But curiously absent were representatives from the United States, probably reflecting the topic du jour: sequestration. Imagine holding a geo-spatial and remote sensing conference and no one from NASA was able to attend?

From an international perspective, we observed tremendous interest from customers looking to gain spectral capability for their manned aircraft and also surprising interest from organizations looking to buy “all-inclusive” UAV configurations that include the Micro-Hyperspec imaging spectrometer, a GPS/INS unit, a lightweight IGARSS 2013 Boothembedded processor, and an suite of application software. This complete airborne package was a big hit at IGARSS because while users have good grasp on the benefits of airborne hyperspectral, they need help making it work in particular application.  Two very nice UAVs on display at IGARSS created a lot of buzz in the Headwall booth. Although Headwall doesn’t make the UAV platform, we make them do some pretty amazing things within the realm of hyperspectral remote sensing. That message came through loud and clear, as our stand at IGARSS was phenomenally busy from the start right through the end.

A bit further up in altitude were visitors interested in hyperspectral remote sensing from space. A major point of interest throughout the conference was a demonstrated need for cost effective, space-qualified hyperspectral sensor payloads.  With most of the world’s planned remote sensing missions being delayed for budget reasons, VNIR (380-1000nm) and SWIR (900-2500nm) space-qualified imagers are hot commodities. This is an area that Headwall Great Ocean Roaddeveloped over the last five years with its own space-qualified sensor payloads.  There was also strong focus from attendees on how satellite collaboration could be established among the world’s most notable remote sensing programs.  Japan’s ALOS-3 (2016 launch?), European ENMAP (2017 launch?), and NASA HYSPIRI mission (2023 launch?) represent three of several.

Even with all the activity at IGARSS, Headwall’s remote sensing team led by Kevin Didona, Principal Engineer at Headwall, also took some hyperspectral scans of rock wall formations at some very scenic places along the Great Ocean Road on the South Coast of Australia.

As Headwall has developed extensive experience in the application of hyperspectral sensors specifically designed for UAVs, please drop us a line or give is a call if we can provide some information to meet the objectives of your remote sensing research.

Email us at Information@HeadwallPhotonics.com

Visit us at www.HeadwallPhotonics.com

Or call us at Tel: +1 978 353 4003


Hyperspectral and Remote Sensing the focus at EARSeL!

 

Headwall's exhibition schedule kicks into high gear this month. First up is our appearance at the 8th Imaging Spectrometry Workshop, sponsored by The European Association of Remote Sensing Laboratories (EARSeL). This event gives visitors the opportunity to understand how hyperspectral imaging can be a valuable scientific tool for the research community. Precision agriculture, mining & minerals, petroleum pipeline surveillance, and disaster mitigation are just a few application areas and more are uncovered all the time as the technology becomes more affordable and easier to use.

EARSeL blog photoHeadwall is seeing a meteoric rise in the use of small and light UAVs for remote sensing activities. SkyJib (from Droidworx) and the Mk II by Winehawk Labs are two such examples, and you’ll see both at EARSeL. The more nimble these hand-launched airframes get, the smaller and lighter the sensors themselves need to be. Headwall’s collaborative engineering approach gives customers a fast path to success with lightweight solutions that also include integrated application software and a GPS/INS. The beauty of Headwall’s Micro Hyperspec sensor is that it is purpose-engineered for flight. Besides being rugged, it also provides outstanding spatial and spectral resolution in the NIR (900nm-1700nm) and VNIR (380nm-1000nm) ranges while also having a very wide field of view. A wide field-of-view means a more efficient the flight path. In other words, the UAVs can cover more territory by collecting precise spectral detail not only directly below but also off to the sides.

While small, hand-launched UAVs are perfect for a wide range of scientific exploration activities, fixed-wing aircraft ranging from the Cessna to the Twin Otter are also used as a platform for hyperspectral sensors. Headwall’s High-Efficiency Hyperspec sensor covers the NIR (900nm - 1700nm) and SWIR   (950nm - 2500nm) spectral ranges. Aberration-corrected and completely athermalized, it provides the highest optical performance and diffraction efficiency of greater than 90%. We’ll be showing this at EARSeL also.

Later in April…beginning on the 3oth actually…Headwall will be at the Defense, Security + Sensing show in Baltimore. We’ll be in Booth 1830 at the Baltimore Convention Center for DSS, which is quickly becoming the go-to show for all things related to surveillance and reconnaissance. While the interest here is largely airborne, visitors also want to know about ground-based and hand-held hyperspectral sensors. Headwall’s flagship hand-held sensor is Hyperspec RECON, which won the R&D100 Award in 2012. This portable instrument covers the VNIR (380nm-1000 nm) spectral range and can render a 6-inch sq. hyperspectral scene at a distance of over a kilometer. Best of all, it’s easy to use and can be ‘tuned’ by loading spectral libraries via an integrated SD slot. Hyperspec RECON represents a very flexible reconnaissance platform that can also be used in a stationary manner (mounted to a mast or a vehicle, for example).

While Hyperspec RECON and its handheld ingenuity is a groundbreaking achievement, many applications need instruments that can either point-and-stare’ or ‘pan-and-tilt.’ Headwall has sensors for both types of deployment that exhibit the very same aberration-corrected concentric imaging performance as their airborne counterparts. Since hyperspectral imaging depends on movement to occur, the instruments are motorized and fully engineered for the tasks they are challenged with.

Headwall will be at several exhibitions and conferences throughout 2013 aside from the two described here. These events will serve as excellent venues as we come out with new products and enhanced versions of existing ones.

 


 

Resource Exploration Using Hyperspectral Imaging

 

Headwall utilizes hyperspectral sensing technology as an essential industrial inspection platform and has made this technology increasingly valuable across a wider spectrum of commercial applications and most notably in the oil & gas industry.  Companies in the petro-chemical industry focus much of their financial capital and effort on efficient pipeline distribution, refinery operations, and environmental monitoring.  Not only for exploration, but also to keep to keep their refining and distribution infrastructure safe.

hyperspectral analysisSo how can hyperspectral sensors help?  The lessons and knowledge gained from the remote sensing applications are directly applicable to the challenges faced by oil & gas companies as very remote and harsh territories are managed for energy production.  The data-rich imagery produced by a airborne and ground-based hyperspectral sensor can provide answers to some of the most pressing questions:

  • Are pipelines being properly monitored for structural integrity and vegetation encroachment?
  • Are pipelines leaking products such as methane?
  • Is there environmental damage that cannot readily be observed?
  • Does a particular area hold exploration value?

In practically every case, these questions are posed with respect to some of the most remote and desolate territory around. The upper reaches of Canada, Siberia, and within the Arctic Circle to name just three.  It’s practically impossible to simply drive over this rugged ice and permafrost terrain, which is why companies in the petro-chemical industry invest so heavily in airborne assets such as fixed-wing aircraft and UAVs as well as invest in satellite-based remote sensing data.

PipelineHyperspectral sensors measure the intensity of solar energy reflected from materials over hundreds of wavelengths from the visible-near infrared (VNIR) to the long wave infrared (LWIR) spectral region. They can record visible light (comprised of relatively short wavelengths such as blue, green, and red) as well as longer, near-infrared, and short wave-infrared light. Reflected light is collected into picture elements (pixels) by flying the imaging sensor over terrain. The reflected visible and infrared light is subdivided into 100 to 200+ discrete wavelength bands within each pixel.

Headwall has developed a leading position in the manufacture and deployment of small, lightweight hyperspectral sensors that are specifically designed for the small, low flying UAVs being deployed. Not only are the sensors small but they generate high resolution spectral and spatial imagery.  The patented, aberration-corrected design of the Micro-Hyperspec sensor allows UAVs to make fewer passes over a certain geographical area while eliminating image aberrations.

Crude oil can be ‘seen’ by hyperspectral sensors operating in the visible/near-infrared spectral bands. A phenomenon known as ‘micro-leakage’ yields hydrocarbon components in the surface soil and water, which the sensors can detect. There is a correlation between ‘micro-leakage’ and the probability of an oil or gas reservoir; detecting the presence of hydrocarbon is a technical means of making that correlation. Doing so from a UAV means a much more efficient collection of useful data as the sensor can be designed to ‘discriminate’ and ‘see’ precisely what geologists are hoping to see based on the spectral signatures of interest.

Disaster mitigationOther useful deployments of hyperspectral include looking at the state of vegetation stress near oil and gas pipelines. With legislation such as California’s “cap & trade” regulations being implemented, managing pipeline content and distribution network integrity carries financial implications for the producers.  With this requirement, the detection of methane from pipeline leaks becomes critical.  With pipelines several thousand miles long, airborne analysis is the only real way to collect actionable data rapidly and with some frequency.

Finally, oil and gas exploration companies are using hyperspectral sensors as a means of environmentally monitoring.  This is very important as environmental changes are often much noticeable utilizing hyperspectral sensor technology to identify spectral anomalies.

In the situation of a spill, hyperspectral sensing can be invaluable in monitoring and prioritizing clean-up efforts. Over the course of time, the sensors can report on trends…both positively and negatively. Again, the ability of hyperspectral sensors to discriminate means more meaningful, actionable data delivered from a cost-effective sensor platform such as Headwall’s Hyperspec imaging sensors.

Mineral mappingWhile the petroleum industry sees value in airborne hyperspectral sensing, so do companies in the minerals/mining industry. Because the cost to explore is prohibitive, innovation at the ‘front end’ means better exploration efficiency. The ability to distill large geographical areas into smaller land packages using airborne hyperspectral sensing means that the more costly assessments can be done where airborne sensing suggests a high probability of success exists.

During the exploration process, hyperspectral sensing can identify the presence of certain minerals such as iron ore and can also ‘grade’ them with a high degree of precision. A weathered environment can also hide the presence of valuable mineral deposits from normal explorative techniques, while hyperspectral sensing can unmask them. This mineral map for the Yeelirrie district of Australia demonstrates the ability of hyperspectral imaging to identify mineral assemblages in the presence of intense weathering. This particular map is indicative of calcrete-hosted Uranium.

 

 

Remote Sensing: All Eyes on Munich

 

The IEEE is an esteemed organization with top-notch events held worldwide. These events draw experts from across industry, government and education.

One of these events is happening next week, in Munich, Germany. The IEEE's International Geoscience and Remote Sensing Symposium (IGARSS) will probably see its biggest attendance ever, as the evolution of unmanned aerial vehicles (UAVs) melds with needs of the remote sensing community. Headwall Photonics will be in booth #18.

IGARSS 2012Much of what scientists want to analyze is best done from above. This holds true for oceanography, atmospheric research, precision agriculture, minerals and mining, and forestry management. Now that commercial UAVs are becoming more affordable and regulations governing their use more ‘mainstream,’ the door is wide open for a fascinating amount of quality research helped along by these small, pilotless aircraft.

Hyperspectral sensors represent a highly desired piece of precision instrumentation carried aloft by UAVs. Why? Because they can extract a tremendous amount of data based on the spectral makeup of what is within the field of view. What the human eye—or even infrared—cannot see, hyperspectral sensors can. Small, lightweight, and extremely precise, Headwall’s Micro Hyperspec is favored for its ability to offer several attractive capabilities. First is its tall slit, which gives the sensor a wide field of view. The wider the field of view, the more precise the hyperspectral data is from a given altitude. Looking down Hyperspectral imaging from UAVsfrom above, UAVs can make fewer passes over a plot of land if the resolution to either side of the flight path is very wide. In short, more territory can be covered in less time.

Another highly desired characteristic is spatial and spectral resolution, which determines how faithful the hyperspectral data is. The beauty of a hyperspectral sensor is that it can delineate what it ‘sees’ with a tremendous degree of resolution. For example, higher resolution can mean the difference between simply distinguishing disease conditions and determining what those diseases are. Or, determining good soil conditions from bad.

While affordable UAVs are all the rage at present, the beauty of hyperspectral imaging is that instruments can be made small and rugged to fit specific payload requirements. 'Size, Weight & Power' (referred to as 'SWaP) describes the continuous desire to make payloads as small, lightweight, and as power-efficient as possible. These characteristics hold true for any airborne vehicle aside from a UAV, whether a fixed-wing aircraft, a high-altitude reconnaissance plane, or a satellite. Headwall Photonics has hyperspectral instruments deployed successfully in all these platforms.

 

 


Hyperspectral Imaging - Next Generation Machine Vision Platform for Food Safety and Quality

 

Bosoon Park, author of this blog entry, works as an Agricultural Engineer on behalf of the USDA in Georgia. He has done extensive research on hyperspectral and Raman imaging as it applies to food inspection and agriculture. Author of numerous published papers on the subject, Bosoon will be co-presenting a discussion on hyperspectral imaging at the annual conference of the American Society of Agricultural and Biological Engineers to be held in Dallas July 30 through August 2.

USDA At USDA, our work revolves around making sure that the foods we harvest and eat are safe, high quality, and healthy. Our mission is twofold: ensuring and improving the safety of food and feed, and ensuring and improving the quality and economic value of food and crops.

There are very important inspection steps between ‘farm’ and ‘fork,and the USDA invests considerable time looking at new technologies that can help. Hyperspectral & Raman imaging (both imaging spectroscopy techniques) can provide valuable inspection data based on the chemical composition of agricultural products that traditional machine vision systems cannot provide. 

During the past decade, USDA has worked with companies such as Headwall Photonics to develop hyperspectral technologies for in-line food safety inspection. Our work focuses on contaminant detection during in-line processing, which the Hyperspec Inspector allows us to do. Our researchers are expanding hyperspectral imaging technology to rapidly detect foodborne pathogens at a microscopic level.  Hyperspectral imaging has tremendous potential for the food industry in terms of safety inspection and quality control by analyzing spatial and spectral characteristics of agricultural products.  We are also exploring handheld hyperspectral instruments fully integrated with operating software for field use.

Raman spectrometers will also detect foodborne pathogens since their scattering phenomena respond very well to particular laser-lighting sources. USDA researchers have proved the concept to identify bacterial species and foodborne bacterial serotypes with surface-enhanced Raman scattering (SERS). This is an emerging area of focused research for improved food safety.

In an effort to educate and inform, several of us from USDA are preparing a short course on ‘hyperspectral imaging’ at the upcoming American Society of Agricultural and Biological Engineers (ASABE) conference July 30-August 2 in Dallas. Thanks to help from Headwall Photonics in commercializing and economizing the technology, we’re able to research and test hyperspectral and Raman instruments so that they can become mainstream across food-processing industries ranging from poultry to specialty crops.

All Posts