Seeing the invisible: The Basics of Hyperspectral Imaging

The Basics of Hyperspectral Imaging

Our human eyes are powerful electromagnetic wave detectors that react at wavelengths between 400 nm (purple) and 700 nm (red). However, the wave spectrum is much larger and invisible to our eyes. Technology today allows us to have special cameras to detect a greater range of frequencies, these are called hyperspectral cameras, capable of obtaining complete information from a band of the spectrum.

Operation: Sensors and many colors

To understand how it works, if we inspect the sensor, we will see that the way it operates is very similar to the camera that our cellphone has, but a little more complex (and expensive).

Like common cameras, a hyperspectral camera is composed of an array of sensors that are sensitive to a certain range of the electromagnetic spectrum, generally the range that comprises the Near Infrared or NIR. Then, each pixel reacts differently to the frequencies it receives. That is, if in a pixel of our cellphone the light is broken down into three RGB levels (Red, Green and Blue), in the case of hyperspectral, a pixel is represented by hundreds of components each corresponding to a different wavelength in that band of the spectrum. 🤯

Know what an object is made of without touching it

Well, we already know a little more about how the sensor works, but what is the physical phenomenon that this camera tries to represent?

It’s like this: When any object in nature is exposed to a source of photons (or light) they react at an atomic level to that source, absorbing, returning or dispersing part of that energy. Taking advantange of this, a hyperspectral camera notices this change in the object when being exposed to light or energy source, and here is the magic. A stone has a completely different reaction to that of a tree leaf, that is, both return different things, or in technical language, they have a different hyperspectral signature, in this way we can know how a material is composed without touching it and only illuminating it (at least, superficially).

The main requirement for proper operation is to illuminate the object of study with enough energy, so a satellite can only capture hyperspectral images during the day, unless it has a mega light bulb.

Finally, it is interesting from an academic and industrial point of view to extract said information for its interpretation and improve decision-making. At Odd Industries we use these images because they are extremely rich in information and have innumerable applications for understanding our natural world.