Point Cloud Data
DEMs were the first widespread mechanisms for
distributing and displaying digital topography.
Point clouds are collections of 3D points located randomly in space. A
sensor emits a pulse of energy and times its return trip (TWTT, two way travel
time). Knowing the
position of the sensor, and the direction in which the pulse was transmitted,
the 3D location of the reflecting surface can be determined. The sensor
can also measure the intensity of the return, to estimate the surface geometry
and material composition of the reflecting surface.
The point cloud can be used directly, or converted to a 2.5D grid, a DTM
or DSM. As a grid it can be smaller in size, more familiar, and easier
to manipulate, than as the point cloud.
|
|
 |
Lidar is now generally spelled using all
lower case letters, and stands for "light
detection and ranging".
Airborne lidars
(laser), fly on a helicopter or small plane,
and now on drones.
These show the ground, vegetation, buildings,
and power lines, and even unwanted objects like
birds.
This shows a 3D depiction of the point cloud.
With computer graphics, the point can be rotated
about three axes, zoomed in and out.
- Overhead view
- Sides of buildings and cliffs generally obstructed
- Fast
- Large footprint (10's cm)
- Longer ranges
- Point density 1-80 points/m²)
|
 |
Terrestrial lidar, from a scanner set on a tripod. This can done
outdoors or in a lab or museum. New technology is expanding lidars to
work underwater, despite the much greater
attenuation of light compared to sound. This slice in the XY plane shows the
locations of the sensor, at the center of each
of bull's eyes, on the deck of a ship in a
museum, because the scanner is limited in how
low it scans and it does not directly below the
tripod.
|
 |
Mobile lidar can be mounted on a
backpack, or on a vehicle (car or train) for
mapping corridors or a drone, with a very large point
density and relatively rapid data collection.
- Good view vertical faces
- Blocked areas
- Relatively slow (at least compared to airborne)
- Small pulse footprint (few cm)
- Shorter ranges (<100 m)
- Higher point density (100s to 10000s of points/m, but variable with range and other factors
|
 |
Scanning Sonar, from a ship or ROV. For really
high resolution results this will be from an ROV
which will slowly survey the object. The
sonar can also be placed on a tripod on the
bottom, and then moved a few times for complete
coverage.
Sound travels at about 1500
m/s, a small fraction of the speed of light.
This shows a slice through the point cloud. |
 |
Photogrammetry. We can now create
point clouds using a series of photographs,
finding matching points, and deriving the point
cloud from that. This is from photographs
of a ship model in a museum. Traditional photogrammery
used high precision cameras and equipment to
manipulate the photographs, and required skilled
operators.
Structure from motion (SfM) is
a range imaging technique; it refers to the
process of estimating three-dimensional
structures from two-dimensional image sequences
which may be coupled with local motion signals.
SfM can use consumer grade cameras and computing
power to rapidly and nearly automatically create
point clouds.
Photography will not
penetrate foliage, so the resulting surface will
be a DSM.
It is
much cheaper to put a good camera on a drone,
compared to a lidar sensor, so this offers a lot
of potential if you can live with the DSM and no
ground points, or acquire the ground points some
other way.
Image from Julia
McKenna. |
Lidar return classification
LAS file format, the
standard for point clouds in the mapping realm
You can also view 2D slices through the point cloud. While not as visually
exciting as the 3D rotating displays, slices allow measurement and may more
clearly reveal relationships.
Point cloud display controls
Problems with point clouds
Other point cloud viewers
Interesting point clouds
last revision 3/28/2019