To search for explosives, chemical warfare agents (CWAs), and toxic industrial chemicals and materials (TICs and TIMs), a capability for standoff/proximal detection and identification of surface contaminants is required for high-throughput, covert screening of people and vehicles. However, current trace detection systems—such as those employed in airports—require direct contact (i.e., swabbing). In addition, these methods are neither covert nor high throughput. In contrast, an ideal standoff detection approach should enable screening that is eye-safe, wide-area, and has a high aerial-coverage rate. It should also have sensitivity and specificity that are comparable with currently available contact sensors. Previously developed standoff (tens of meters) and/or proximal (about 1m) trace surface-residue detectors are based on optical methods. These standoff optical techniques include Raman,1 quantum-cascade-laser-based (QCL-based) reflectance,2 and QCL-based photothermal3 spectroscopy, but they are associated with a number of problems. For instance, they have limited sensitivity and specificity levels (particularly in real-life settings). The sensitivity required to detect trace levels of residues at standoff ranges is also typically at odds with high area-coverage rates. In addition, very bright, directional, non-eye-safe, and almost exclusively coherent illumination sources are required to achieve high sensitivity. This use of coherent illumination, however, typically results in a speckle-noise limit from real-world surfaces. Indeed, the speckle-noise limit is often much worse than the detector-noise limit. Lastly, the extraction of very weak target signatures from a complex scene requires sophisticated detection algorithms that are capable of dynamically compensating for real-world clutter. In our work at Physical Sciences Inc. (PSI), we have successfully addressed all the shortcomings of current trace surface residue detectors with a new sensor platform that is based on tunable QCLs. Our long-wave IR (LWIR) QCL-based surface contaminant detector platform has been developed as part of the Joint Project Manager Next-Generation Chemical Detector program.4 We have developed our platform (illustrated in Figure 1) with several key attributes. For instance, we employ a bistatic, integrated transmitter/receiver within a monolithic package. This features a spectrally tunable external cavity (EC) QCL transmitter, which is used to flood-illuminate a 5 × 5cm region of interest at 1m standoff with eye-safe irradiance levels, as well as a broadband, uncooled microbolometer focal-plane-array imaging receiver. Another key feature is the reflective calibration target we use for illumination/collection and spatial-profile normalization. We also use an intelligent speckle-reduction technique, and a robust detection algorithm that is based on the adaptive cosine estimator (ACE).5, 6 With our platform, LWIR reflectance spectroscopy is performed at proximal ranges. During this process, the eye-safe QCL is used to actively probe—with the highest LWIR spectral brightness available—fundamental absorption features of optically thick and thin materials (including solid- and liquid-phase CWAs, TICs, TIMs, and explosives). In particular, we use a Daylight Solutions MIRcat EC-QCL that covers the 900–1300cm−1 spectral range, with a minimum power of 100mW. We chose to use this QCL in concert with an FLIR Tau 336 receiver so that we can achieve a detector-noise-limited signal-to-noise ratio (SNR) of 5 (relative to the discriminant reflectivity modulation that is associated with relevant target surface loading levels). This SNR of 5 is the minimum requirement for the ACE algorithm. With our variant of the ACE algorithm, we enable contaminant detection and identification in highly cluttered environments and in the SNR regime that can be achieved with the sensor for surfaces of interest. We use speckle mitigation, via temporal and spatial diversity, to maintain the platform's sensitivity (which is limited by the QCL's spectral power and the receiver noise). Spatial diversity is the ratio of the resolving power of the transmitter aperture to that of the receiver. High spatial diversity therefore results from averaging many speckle cells with the receiver. Temporal diversity refers to the number of independent speckle patterns that are generated during the integration of the receiver.7 We achieve high temporal diversity by using a spinning reflective diffuser at the focal point of an off-axis parabola transmitting optic. In our system, the spatial and temporal diversities work together to produce a speckle-noise-limited SNR of ≥ 20 at all wavelengths, which corresponds to a noise-equivalent-reflectivity (NEρ) of 5% for full-speckle-forming surfaces. With our detector platform we can thus achieve a speckle-mitigated and detector-noise-limited performance. The sensitivity is sufficient for accurate detection and discrimination, regardless of surface coverage morphology and underlying surface reflectivity. We have performed developmental testing to demonstrate the capabilities (including the detection algorithms) of our sensor system. We prepared test substrates by fabricating different size wells in white-painted aluminum, chemical agent resistant coated (CARC) steel—see Figure 2(a)—and high-density polyethylene (HDPE), to simulate a range of droplet sizes. We then filled the wells sequentially with four CWA simulants, including triethyl phosphate—TEP—which is a VX (i.e., Venomous Agent X) simulant. We acquired spectral images of these targets and converted them to reflectivity spectral images with a spectral image of the calibration target. Finally, we applied the ACE detection algorithm to four target compounds (reflection and absorption) and to polystyrene. Our detection results for the three substrate materials and TEP, where the six largest wells were filled—as indicated by the arrows in Figure 2(a)—are shown in Figure 2(b–d). The detection output for the CARC steel, with detection on five of the six wells, is shown in Figure 2(b). Some optically thick and thin bleed over can also be seen in this image. As expected, the filled wells generated a specular reflection from the liquid, whereas the bleed over generated a diffuse reflection off the TEP. The diffuse reflection was from the TEP, conforming to the CARC substrate (optically thick) and/or the CARC surface itself (optically thin). The specular reflection from the meniscus of the liquid was significantly smaller than the well and created a sub-pixel detection scenario. The detection results on white-painted aluminum are shown in Figure 2(c). In this image, all the wells were detected and included the bleed over. Lastly, the results on HDPE are shown in Figure 2(d), where five of the six wells were properly detected. In summary, we have designed and tested a spectrally tunable QCL-based detection platform for proximal solid and liquid CWA detection. For our sensor, we employ eye-safe flood-illumination with an imaging receiver. In addition—through speckle reduction—we can achieve NEρ of better than 5% with our sensor, which enables detection at the 5–10μg/cm2 surface-loading level. With the results of our developmental testing, we demonstrated CWA simulant detection in both optically thick and thin regimes for both diffuse and specular reflection scenarios that exercised the full capability of our sensor platform. We are currently applying our new understanding of speckle mitigation, and algorithms for detection in a cluttered environment, to new surface contaminant detection programs and applications. Specifically, we are developing approaches for further speckle reduction and class-based target screening for false alarm mitigation. This work was supported by Chemring Detection Systems and PSI internal research and development funding.
News Article | April 7, 2016
Over the last four months of 2015, US federal agencies have secretly flown spy planes above American cities more than 3,500 times. That's according to data from the aircraft tracking site FlightRadar24 collected by BuzzFeed News to map out the clandestine surveillance flights, which came to prominence last summer after it was discovered that the FBI had flown light aircraft in circles over the Black Lives Matter protests in Baltimore, MD and Ferguson, MO. The data offers the most comprehensive portrait yet of hidden aerial surveillance operations run by the FBI and Department of Homeland Security, which routinely patrol the skies above most major cities looking for terrorists, human traffickers and other “serious criminals.” Well, except maybe on weekends. The telltale sign of these secret flights is thick circular lines indicating aircraft—often single-engine planes registered to fake companies with names like FVX Research and National Aircraft Leasing Corp—flying around the same area over and over for hours at a time. These patterns can be found by searching public FlightRadar data (which tracks aircraft positions by recording their transponder broadcasts) against a list of known government front companies and aircraft registration numbers. But the circular patterns practically disappear on Saturdays, Sundays and federal holidays, when the average flight time drops off more than 70 percent. That suggests that whatever crucial missions the Feds are flying, they're apparently not crucial enough to warrant denying pilots their weekend leisure time. In response to initial reports last summer, the FBI claimed that the planes are used “to follow terrorists, spies, and serious criminals,” and that they are “not equipped, designed, or used for bulk collection activities or mass surveillance.” The Bureau did admit that the planes are sometimes equipped with cellphone mass-surveillance devices better known as Stingrays, but that this is rare and only occurs “with a court order, or under exigent circumstances such as a hostage situation.” One of the most striking examples comes from the day of the mass-shooting in San Bernardino, CA last December. The flight data shows that within 90 minutes, two planes—one flown by the FBI and the other by Homeland Security—were circling the site of the attack. The FBI plane later started flying around the home of the two shooters, Syed Farook and Tashfeen Malik. The following day, it was joined by two more FBI planes which began making circles around the nearby mosque that Farook attended, some for as many as 3 hours at a time. The planes stopped over the weekend, resuming their flight pattern the following Monday. The data shows that the FBI and DHS have also routinely flown planes over Muslim areas in other cities including San Francisco and Minneapolis. The planes fly low at altitudes between 4,000 and 6,000 feet, and are equipped with high-resolution FLIR Talon camera payloads and mufflers to reduce engine noise for stealth. It’s not clear why the planes are grounded on weekends, although it could be that due to their secretive nature very few pilots are cleared to operate these flights. Another possibility is that the flights have more ground support staff available during the week. Alternatively, it could be that these missions are a lot more quotidian and indiscriminate than the FBI and DHS are letting on.
News Article | March 16, 2016
News Article | December 11, 2015
Popular drone manufacturer DJI has announced a partnership with FLIR Systems to build a thermal imaging camera for drones. The camera will be built for the Inspire1 and Matrice 100 drone series', and will be very helpful for things like measuring a home's thermal efficiency and spotting fires. DJI has fast become the most-used drone company for people like cinematographers, who are looking for new ways to capture footage from a height, although the company has recently been trying to find ways to help other industries and causes as well. For example, it recently launched a drone model that has pesticide-spraying capabilities. The company has also published a video showing how the drone could be used by firefighters to spot a blaze and see where a fire originated, the direction that it's spreading in, and whether or not a building is stable. The Zenmuse XT Thermal Camera will come in two resolutions, either 640 x 512 or 336 x 256, and will be controllable from the DJI app. Not only that, but the thermal imaging camera will be extremely sensitive, with FLIR saying that it will be able to measure differences to the tenth of a degree. The two companies haven't yet announced a price for the drones, although FLIR does already make a similar camera that starts at $1,500. The FLIR One, which is a smartphone sensor, only costs $249. It will be interesting to see how the price will affect who can and cannot take advantage of the new camera. Either way, we'll find out when the camera is released in the first quarter of 2016.
News Article | June 24, 2011
Stratford-upon-Avon is best known as the birthplace of playwright William Shakespeare, whose works have been told and retold for centuries. But this scenic town 100 miles northwest of London could be the sight of another tale worth retelling, that of Simon Wheatcroft. At noon local time, 90 people took off at the starting line of the Cotswold 100, a grueling 100-mile ultramarathon that sneaks its way north through the British countryside before circling back toward south Warwickshire. Wheatcroft, however, won’t be able to soak in all the scenery as he slogs his way along the course. That’s not because his attention will be fixated on the task on the hand — finishing the race in the maximum 30-hour time limit — although it certainly will be. You see, Wheatcroft is legally blind and has been so for the last 11 years. He has a degenerative eye disease called retinitis pigmentosa, which directly affects the retina, so his eyes can’t properly convert visual cues to nerve signals that get processed in the brain. While somewhat genetic, the condition affects roughly one in 4,000 people in the United States, according to the National Institutes of Health. There is no cure or effective treatment. For the last few months, he’s been dutifully keeping Wired.com readers apprised of his ever-increasing stamina, how his nutrition regimen has evolved, the hunt for a new team that will help pace him through the event. Finally, the time has come to run the race, which won’t be easy. Yes, the course starts off at a manageable 112 feet above sea level, but it’ll go as high as 1,037 feet by mile 18, a 925-foot elevation differential. That’s the high point of the race, and from there the hills will be frequent, although not too demanding. The last major obstacle will come just after Wheatcroft passes mile 80 — his own Heartbreak Hill, if you will — as he’ll experience a 400-foot-plus incline in less than three miles. Today’s weather calls for temperatures reaching 66 degrees under partly cloudy skies, although rain is slated for Saturday, when contestants will start crossing the finish line. Throughout the entire training progress, Wheatcroft has dedicated much of his time to sharing his story with readers around the world, both on Wired.com and his personal sites, AndAdapt.com and Blind100.com. That’s not changing today, even as he’s in the middle of running. In addition to sending out pictures and video updates on his @moochoo Twitter account, Wheatcroft will have reader questions relayed to him by his team on the ground. Just include the hashtag #blind100 and you might just get an answer back from Wheatcroft in real-time. You can also follow along, graphically speaking, since Wheatcroft will be using RunKeeper’s livecasting service. Here, you’ll find real-time updates on how Wheatcroft is progressing along the route, as well as a quick look at whatever challenges elevation may play in his remaining miles. Wheatcroft’s story has inspired many readers since he started blogging and tweeting, but it’s all come down to today, just a few more hills to climb. And when Simon crosses the finish line Saturday, that will certainly be a sight to behold.