Introduction to Hyperspectral Imaging System

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
950 1450 1950
Hyperspectral imaging:
Introduction
Hanim Zuhrotul Amanah, S.T.P., M.P., Ph.D.
Department of Agricultural and Biosystem Engineering
Faculty of Agricultural Technology
Universitas Gadjah Mada

References
You may use other references. More references, more better!

Spectroscopy: Light versus matter
We have an apple and we have a light source to illuminate our apple.
Some light will be
bounced back
(reflected) by the
apple’s surface.
Incident beam
Some light will be
absorbed by the
apple.
Due to apple’s
texture, the light
can also be
scattered.
Incident beam
Incident beam
Apple surface
© Copyright Hanim Zuhrotul Amanah, Ph.D. 2024

So, what is going on behind those events?
Reflection
When an incidental beam
passes through a medium
which differ in refractive index
(RI).
If the medium has higher
refractive index, the
fraction of reflected
radiation will be higher.
θ1
θ2
M1
M2
𝑛2 =
sin 𝜃1 𝑎𝑖𝑟
sin 𝜃2
Normal
Ex.: water (RI = 1.33)* has low RI compared to glass
(RI = 1.50)* thus, light will be more reflected by glass.
*Using yellow light. Refraction of light passing from less (M1)
to more dense (M2) medium
Refractive index

So, what is going on behind those events?
Absorption
When the light is not 100%
reflected.
Io If
As illustrated, there is a reduction of the
light intensity. It is because the apple
absorbs some light.
It is slightly different from reflection event, in absorption,
we need to see at atomic level.
P0
P1
Pn
Incident beam
Electron
𝐸𝑛 = ℎ𝜈𝑛 = ℎ𝑐/𝜆𝑛
Some of the energy from the incident beam is
being used by the electrons to move from the
ground state (lower) to higher energy.
The remaining light intensity (If) is called as transmission.

Spectroscopy: Measurement
• Well, spectroscopy is the study of the interaction between light and matters (as described
above).
• Spectrometer is an instrument to measure the value of reflectance, absorbance, etc., as a
function of wavelength.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
950 1450 1950
Illumination
source
Wavelength
selector
Sample Detector Spectrum
Typical spectrometer system. The position and illumination source can be customized depends on the needs.

• Usually the spectrum measurement is conditioned based on the regions of electromagnetic spectrum.
• In this case, this class will be focusing on the spectral regions spanning from visible to infrared.

There are two way to capture the spectrum profiles:
Point-scan Line-scan

Point-scan
In point-scan method, the spectrum is collected in
a single pixel of each sample.
This method is the common method for
any spectroscopy techniques.
It can also be

Line-scan
• In line-scan method, the
spectrum is collected
from the entire surface of
the sample, allowing the
measurement at all pixel.
• Thus, the results will be
in the form of image,
where the reflectance
value is presented in
each pixel.
This technique is well-
known as hyperspectral
imaging.

Hyperspectral imaging
Hyperspectral imaging or imaging spectroscopy or spectral imager
is an instrument to measure reflectance radiations of a sample within
field of view (FOV), enables the user collecting a simultaneous
information at one measurement.
Type of data
Spatial
Spectral

Hyperspectral imaging: HSI vs. RGB
Hyperspectral imaging is a combination of spectroscopy + imaging, BUT, it is unlike our phone
camera or your DSLR camera.
Image resulted from our camera is only
composed by three channels, namely red
(R), green (G), and blue (B).

Let’s capture an apple image using your phone (or download through internet) and read using
MATLAB software.
Simple MATLAB code

Let’s capture an apple image using your phone (or download through internet) and read using
MATLAB software.
Using the “imread” and “imshow”
syntax, we can see that the image
only contains 3 channels.
Any idea about the RGB value between the apple and its leaf?

• However, different with common camera, the image resulted from hyperspectral imaging is in
three-dimensional (3D-hypercube), making it have a big size of image (up to 100 MB).
3D-hypercube image
Single pixel
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
950 1150 1350 1550 1750 1950
Intensity
(a.u.)
Wavelength (nm)
Spatial, x
Spatial,
y
Composed as two dimensional image corresponding at
each band, concatenate into a 3D image.

Various types of HSI
Hyperspectral image can be measured in various spectral ranges, as described in the previous
slide.
Hyperspectral
imaging
Visible-near infrared (Vis/NIR) HSI.
Shortwave infrared (SWIR) HSI.
Raman HSI.
Fluorescence HSI.
Different on the light
source and type of
detector.

Light source and detector: Brief information
Continuum sources: emit radiation that
changes in intensity smoothly as a function
of wavelength.
Line sources: emit a limited number of
lines, or bands of radiation.
Each of which spans a limited range of
wavelengths.
So far, HSI only employs an
LED, Tungsten, and lasers as
the illumination source.

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
950 1450 1950
Thank you

Introduction to Hyperspectral Imaging System

More Related Content

Similar to Introduction to Hyperspectral Imaging System (20)

More from HanimZuhrotulAmanah1 (6)

Recently uploaded (20)

Introduction to Hyperspectral Imaging System