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.
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.
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?