RANDOM POINTS
LEWIS GRAHAM
The Advent of 3D Color?
O
ne of the relatively new
advances in commercial
LIDAR technology is multibeam systems with different wavelengths
for each beam (unlike multibeam with the
same wavelength used simply to increase
effective pulse rate). These so-called
“multispectral” LIDARs are becoming
quite common in the bathymetric arena
where infrared is needed to detect the
water surface and green is needed for
water column penetration.
In 2014, Optech introduced a new
system (“Titan”) specifically aimed at
extending the multispectral notion
beyond simple infrared and green. Are
we at the onset of multispectral active
sensors? Let’s casually explore what is
meant by multispectral when it comes
to LIDARs.
Light is part of what is called the
electromagnetic energy (EM) spectrum
or electromagnetic radiation (EMR)
spectrum. You can think of this energy
as having some properties similar to
an ocean wave. The distance between
the crests of the wave is called the
“wavelength.” Interestingly, the energy of
EMR is related to the wavelength in an
inverse way. The shorter the wavelength,
the higher the energy. This sort of makes
sense because more cycles of waves can
be packed into the same space if the
distance between the crests is shorter.
Electromagnetic energy can be of
any wavelength ranging from virtually
infinity (very long wavelength) to very
Figure 1: The Electromagnetic (EM) Spectrum
“ ost cameras use what are called “passive”
M
sensors. They rely on the scene being illuminated
by some external source of radiation.
short wavelengths. The light that we
see is a small range of wavelengths
that extend from the longest we see
(deep red) to the shortest we can
discern (violet). Above what we can
see are ultraviolet light, x-rays and so
forth. Just below what we can see is
infrared. Radio, gamma, microwave
and other wavelengths are in the
EM spectrum but far outside of the
ranges we perceive with our eyes. The
electromagnetic spectrum is depicted
in Figure 1.
”
“White” (or “visible”) light contains a
mixture of a wide range of wavelengths
in the visible region. You may recall
experiments with a prism, separating
out the color components from white
light. To me it is rather amazing
that our eyes and visual cortex can
distinguish these individual colors.
We actually have a very sensitive
“spectrum analyzer“ built in to our eyes
and brains! Creatures that cannot see
color (rare) do not have this ability to
separate out wavelengths. They just
Displayed with permission • LiDAR Magazine • Vol. 6 No. 5 • Copyright 2016 Spatial Media • www.lidarmag.com
perceive them as levels of brightness;
e.g. a grey tone image.
Charged Coupled Devices (CCDs)
and Complementary Metal Oxide
Semiconductor (CMOS) image collectors
used in digital cameras have a wide range
of spectral response (they respond to
light from the near infrared to the violet).
For color cameras, a color lens is placed
over the CCD elements, allowing only a
narrow range of wavelength to impinge
upon the sensor. In standard cameras,
these are red, green and blue filters.
Most cameras use what are called
“passive” sensors. They rely on the scene
being illuminated by some external
source of radiation. For example, an
aerial camera relies on the sun illuminating the scene. You can think of broad
spectrum (e.g. white) light being emitted
by the sun, illuminating the surface
of the earth, reflecting to an airborne
camera lens and then being filtered
down to red, green, blue (and sometimes
infrared) by the color filters on the CCD.
This filtering of multiple “spectral” bands
of light is what we call “multispectral”
imaging (“more than one color”). Note
that if the sun (or some other source of
illumination) is not present, we cannot
image with the sensor.
A laser scanner, on the other hand, is
an active sensor. It illuminates the scene
with laser light and then detects the
reflected light. Thus it can work night
or day since it is providing its own light
source. For a variety of reasons, we use a
laser as the source of illumination. One
of the significant properties of a laser
is that the emitted light is of a single,
very narrow wavelength. Since a specific
color implies a specific wavelength, this
means a laser is one pure color. Most
topographic lasers use a wavelength
of 1,054 nanometers which is in the
Figure 2: Titan Wavelengths
infrared region. This particular wavelength is used because it is “eye safe.”
We can visualize the intensity return
from a laser scanner. This visualization
looks like a black and white image (well,
really a grey scale image). It is actually
infrared but we usually simply view it as
a grey scale.
Now imagine that you could add a
green laser and a blue laser to the mix.
Then the return would be somewhat
similar to that imaged by a Red-GreenBlue camera. However, the laser is an
active sensor, illuminating the scene.
This would mean that you could do this
RGB illumination at night and effectively
have a night time camera. We could
think of this as active, 3D color imaging.
The wavelengths of the Optech Titan
are depicted in Figure 2. As previously
noted, it uses three lasers, each having a
different wavelength. Their new system
can be called “multispectral” because
it is illuminating and detecting with
more than one wavelength but it does
not necessarily mean that what you
would image would resemble a ‘normal’
RGB scene. For example, even though
depicted in yellow, the intermediate
wavelength beam at 1,064 nanometers is
actually in the infrared band.
Judson Thomas performed extensive
experimentation with classifying
Titan data using conventional tools
aimed at image classification (“Terrain
Classification using Multi-Wavelength
LIDAR Data”, September 2015). While
his results look promising, it is obvious
that new software algorithms will need
to be devised for dealing with these
very narrow and widely spaced spectral
channels.
These are fascinating developments
that will hopefully lead to an entirely
new way to do spectral modeling.
Perhaps a future development will be a
tunable laser that can be set to mission
specific bands. It will be an interesting
progression to follow.
Lewis Graham is the President and CTO
of GeoCue Corporation. GeoCue is North
America’s largest supplier of LIDAR production
and workflow tools and consulting services for
airborne and mobile laser scanning.
Displayed with permission • LiDAR Magazine • Vol. 6 No. 5 • Copyright 2016 Spatial Media • www.lidarmag.com