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.
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.
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.
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).