Hyperspectral Imaging

A breakthrough in spectroscopic technology

Luca de Leonardis
What is Hyperspectral Imaging
Hyperspectral imaging, sometimes referred to as the imaging spectroscopy, is a
technology that combines the world of digital imaging with spectroscopy. Hyperspectral
imaging is done using hyperspectral cameras which capture images the same way regular
cameras do but collect far more information than regular cameras do. Each pixel on the sensor
of a hyperspectral camera captures a full spectral trace of the incident light. This generates
what’s known as a cube of information. The data is described this way because each pixel can
have hundreds of data points, each corresponding to a wavelength intensity. This technology is
constantly improving and new applications for this technology are arising on a daily basis.

How Hyperspectral Cameras Work

Snapshot imaging
Snapshot imaging, sometimes referred to as non-scanning imaging, is a hyperspectral
imaging technique that collects the entirety of the information at once. In other words, the
sensor can capture the entire data cube in one detector integration time. There are several
benefits of utilizing snapshot imaging, one of which is the lack of moving parts. Moving parts in
imaging can be quite detrimental to the quality of the image as it can cause artifacts, blur,
vibrations, etc. Another reason that snapshot scanning is used is the decreased acquisition
time. Some other hyperspectral sensors can have better resolution but take much longer to
capture the information.
Spectral scanning
Spectral scanning is a method where a 2-
dimensional image is captured with a narrow
bandpass filter in front of the sensor. The filter is
then moved, and another is placed in front of the
sensor. This is repeated several times and all of the
2D images are overlaid to generate a data cube. The
benefit of using this technique is that the desired
spectral bands can be selected for each application.
This does make the device very versatile but does
have some shortcomings. Because the exchange of
these filters is required to produce the data cube,
the device does take time to gather all the data. The
long acquisition time means that the object being
Figure 1: Sensor acquisition methods [3]
imaged needs to be very still and any movement can
cause spectral smearing (the same effect as a long exposure image moving during capture, it
blurs the image).

Spatial scanning
Spatial scanning is similar to spectral scanning just the opposite. A diffraction grating is
scanned across the sensor using a slit. The scene being imaged is essentially sliced and imaged
in slices eventually forming the data cube. There are a few drawbacks with using this technique;
firstly, push broom scanning has the same issues as push broom scanning where the object
needs to stay very still. Secondly the moving parts can cause vibration or artifacts in the final
image. These cameras need to be mounted very securely on a heavy stand because even the
slightest movement can misalign the slices, making the data cube unreadable.
Spatio-spectral scanning
Spatio-spectral scanning sensors are the newest technology being employed. The sensor
was first introduced in 2014. This technology works by incorporating the two principles of
spatial scanning and spectral scanning. The sensor collects information by projecting separated
light onto the sensor and moving the entire sensor so that as the camera pans the room or
scene every part of the scene eventually is captured across the whole sensor. By incorporating
the two concepts, some of the disadvantages of each are alleviated resulting in a smoother
imaging system. This is a very new technique and still is being experimented with.

Applications
Astronomy
Hyperspectral imaging is commonly used to identify a spatially resolved spectral image.
The spectrum very commonly used to determine what things are in space. Having a spectral
trace for every pixel in the image is very useful. In astronomy, this technique is called integral
field spectroscopy. These systems are found on many of the largest telescopes in the world
including VLT as well as the Chandra X – ray Observatory.

Mineralogy
Hyperspectral imaging is very commonly used in mineralogy to identify the composition
of some rocks. Hyperspectral cameras are used to identify minerals like silica, calcite and
garnet. These minerals tend to have the most recognizable reflected spectral signatures when
illuminated with long wave infrared radiation. These cameras have even been mounted on
drones and sent around rocky areas to try and find veins of these minerals. Using these
techniques has proven to conserve the environment because a more targeted mining strategy
is employed.

Figure 2:A set of stones is scanned with a Specim LWIR-C imager in the thermal infrared range from 7.7 μm to 12.4 μm. The
quartz and feldspar spectra are clearly recognizable [4]

Conclusion
Hyperspectral imaging is still a very new imaging technique. New applications for this
technology are being discovered on a daily basis. The possibilities hyperspectral cameras
introduce are very extensive, it’s even being used to detect emotions like happiness!
Works Cited

[1] Nikon, "Microscopy U - Spectral Imaging and Linear Unmixing," Nikon, 2019. [Online].
Available: https://www.microscopyu.com/techniques/confocal/spectral-imaging-and-
linear-unmixing. [Accessed 15 04 2019].

[2] L. Guolan and F. Baowei, "Medical hyperspectral imaging," Journal of Biomedical Optics,
vol. 19, no. 1, 2014.

[3] S. Grusche, Artist, Acouisition techniques for hyperspectral imaging visualized as sections of
a data cube. [Art]. 2014.

[4] H. Holma, Artist, A set of stones is scanned with a Specim LWIR-C imager in the thermal
infrared range from 7.7 μm to 12.4 μm. The quartz and feldspar spectra are clearly
recognizable.. [Art]. Thermische Hyperspektralbildgebung im langwelligen Infrarot, 2011.