View
134
Download
0
Category
Preview:
Citation preview
William Barragán Zaque
Ingeniero Catastral y Geodesta
Especialista en Sistemas de Información Geografica
Msc. Photogrammetry and Geoinformatics
Director Tecnología en Cartografia
Universidad de Cundinamarca
wbarragan@unicundi.edu.co
Abd-el-Hamed Nabial Ibrahim
NATIONAL AUTHORITY FOR REMOTE SENSING
AND SPACE SCIENCES (NARRS), cairo
msc Education- nttti chenai - India
Msc. Photogrammetry and Geoinformatics
Nabila_gis@yahoo.com
Airborne Laser Scanning
Resumen Este artículo presenta algunas características y aplicaciones del sistema de escaneo láser. El Airborne Laser Scanning es un sistema relativamente nuevo que permite la generación de cartografía utilizando escaneo del terreno, donde es posible diferenciar el modelo digital de superficie del modelo digital del terreno. Se presenta adicionalmente el funcionamiento general del sistema, dando sus principales ventajas en la extracción automática de edificaciones.
KEY WORDS: Laser Scanning, Reconstruction, Level-of-Detail, LiDAR, Image processing.
Introduction
Light Detection And Ranging (LIDAR), laser scanning, and laser
altimetry are terms used to describe technology of reconstructing earth
surface or objects on the ground using laser beams from the air. Laser
scanning is widely used in terrestrial and airborne applications to
- 1 -
reconstruct surfaces with a high level of detail. Since the basic compo-
nents of airborne laser scanning and working principles of airborne laser
scanning are well discussed in many literatures,1
Laser Scanning
Light amplification through stimulated emissions of radiation is known as
LASER. Laser beam is used to determine the range between source and
object.
When the purpose of laser ranging is to reconstruct an object
surface, laser beam can be scanned on the object. This group of
ranging points is called point cloud and this scanning technique is called
laser scanning
If the beam emitting device is fixed on the earth to reconstruct an
object, it is called terrestrial laser scanning.
If the laser device is fixed on an airborne vehicle and earth surface
is scanned, it is called airborne laser scanning.
The main advantage of ALS data is that it does not depend on weather
conditions, therefore output is highly accurate compared to traditional
photogrammetry.
1 Schenk T., 2004, Airborne Laser Scanning, Lecture notes (unpublished), Stuttgart University of Applied sciences
- 2 -
Figure 1: Point Cloud
Measuring Principle
Laser has been used in geospatial sciences for several years. Laser
equipment basically consists on an emitting diode that produces a light
source at a very specific frequency. The signal is sent toward earth
where it is reflected back towards the sensor platform. Then, a receiver
device captures the returning pulse signal. By measuring the time lapse
between the sent and received signals, the distance to surface can
be measured.
Pulses and Returns
Laser pulses transmitted towards earth are reflected, absorbed and
scattered based upon the surface characteristics. Reflected pulses are
received by a receiver device in ALS system. While pulses pass through
vegetation, some particles of the laser beam are reflected back from
- 3 -
D = v. t / 2
VA=D cos αA
Where
c – velocity of light
t – Time lapse
VA-Elevation of point
αA -angle from the nadir
Figure 2: Measuring principle
the branches of tree, while the rest of them are reflected from earth
surface. The reflections coming from the tree branches reach the
receiver first and the reflections from the ground reach it last. These first
and last reflections are called first and last pulses respectively. There are
many possible reflections, but in research only first and last pulses are usually
used. Due multi-reflectivity nature of laser beams, different forestry
applications have been possible.
- 4 -
Classification of ALS data:
Raw ALS data point cloud is both dense and noisy, therefore
noises in the ALS data have to be filtered out, there are several
methods implemented based upon the nature of noises.
Classification of ALS is the next step in ALS data processing.
Classification is the process of segmenting the point cloud into
different classes like ground, vegetation, building etc.
ALS data is normally processed as a point cloud data. In certain
applications, ALS data is rasterized using elevation as intensity
values. Rasterized ALS data is called ALS image or LIDAR image.
Sharp break line information can be retrieved only from the
vector data such as point cloud. The following is a general
classification implemented in ALS data.
- 5 -
Figure 3: First pulse and last pulse
Ground point classification:
Ground points are the ALS points lying on the bare earth surface.
These points create a digital terrain model (DTM). There have
been many techniques invented by re- searchers to perform ALS
data ground classification automatically. This process of segregating
ground points from point cloud can be achieved by local slope of
the terrain or statistical methods. Physical objects like buildings,
towers, vehicles need to be removed. Also, points lying on
vegetation are filtered out. Due to the density and accuracy of ALS
data high accuracy contours like 1 foot, 2 feet, 5 feet can be
generated.2
Vegetation classification
Vegetation classification is a process of segregating vegetation
points from the point cloud. The vegetation boundary can also
be extracted from the classified vegetation points. Multiple returns
from laser pulses enable the researchers to classify the vegetation
points easily.
Laser scanning has vast potential for the direct measurement and
2 Schenk T., Digital Photogrammetry, TerraScience, Laurelville, OH 2001
- 6 -
Figure 4: Classification (a) Raw ALS data (b) Ground points ALS data
Figure 5: Raw ALS triangulated points
estimation of several key forest characteristics. The direct
measurements can be canopy height, sub canopy topography, and
the vertical distribution of intercepted surfaces between the
canopy top and the ground. Other forest structural
characteristics such as aboveground bio- mass can also be
modeled or inferred from these direct measurements.
Building points classification:
Digital surface model (DSM) from the point cloud allows
extracting the features like buildings, roads, and other physical
objects automatically. Building planes can be detected using
mathematical characteristics like surface normal, curvature etc.
Extraction of planar building planes and curved planes are possible
from the point cloud. But the detection of building outline has been
a difficult task unless the building footprint is provided.
- 7 -
Miscellaneous object classification
Other than ground, vegetation and buildings features like
transmission lines, roadways can also be extracted. Especially with
the high density ALS data power lines can be modelled easily.
Potentials for 3D building extraction
ALS technology clearly shows its potential in building extraction, the
following are con- sidered as the potentials for extracting buildings
from the ALS data.
Computational geometrical algorithms can be adopted to
detect the shapes of the building roof, walls and other
prominent structures.
High ALS point density is another factor helps to reconstruct
the building mod- els accurately.
Vertical accuracy can be reached to 5cm in ALS.DSM can
also be extracted from the aerial images using image matching
techniques, but 5 cm or less accu- racy is highly difficult in digital
photogrammetry.
ALS is noise free when compared to the aerial images for building extraction.
If the ALS is done in low altitude, building walls are likely to be
hit by the laser pulses. This will allow reconstructing the walls of
buildings.
- 8 -
References
Alharthy Abdullatif and Bethel James S. (2002). Heuristic filtering and 3D feature extraction from lidar data. In ISPRS Commission III, Symposium 2002 September9 - 13, Graz, Austria.
KRAUS, L. Photogrammetry. Vol. I: Fundamental and Standard Processes. 4ª Edición. Vol. II: Advanced Methods and Applications. 4ª Edición. Ed. Dümmler, Colonia, Alemania. 1992, 1997. 397 p. (vol. I), 466 p. (vol. II).
Schenk T., 2004, Airborne Laser Scanning, Lecture notes (unpublished), StuttgartUniversity of Applied sciences.
Schenk T.,(1999)Digital Photogrammetry, TerraScience, Laurelville, OH 2001
Sohn G and Dowman, I. (2003). Building Extraction Using LIDAR DEMs and IKONOS Images, ISPRS proceedings, Volume XXXIV, PART 3/W13. Dresden, Germany,8-10 October.
- 9 -
Recommended