ANALYSIS OF THE BACKSCATTERED ENERGY
IN TERRESTRIAL LASER SCANNING DATA
Norbert Pfeifer a ’ *, Bernhard Hòfle a ’ b , Christian Briese \ Martin Rutzinger c , Alexander Haring a ’ b
a Institute of Photogrammetry and Remote Sensing, Vienna University of Technology, 1040 Wien, Austria
b Christian Doppler Laboratory “Spatial Data from Laser Scanning and Remote Sensing”
c alpS-Centre for Natural Hazard Management, Innsbruck and Institute of Geography, Innsbruck University, Austria -
(np, bh, cb, ah)@ipf.tuwien.ac.at, martin.rutzinger@uibk.ac.at
KEY WORDS: Terrestrial Laser Scanning, Calibration, Radiometry, Intensity
ABSTRACT:
Terrestrial laser scanning provides a point cloud, but usually also the “intensity” values are available. These values are mainly
influenced by the distance from sensor to object and by the object’s reflection properties. We demonstrate that it is possible to
retrieve these reflection properties from the observed range and the intensity value. An experiment with targets of known reflectivity
behaviour is described. Retrieving object reflectivity is also demonstrated for these targets in another experiment, which was not
used to determine the functional relationship between range, reflectivity, and intensity. The Lidar equation describes the received
optical power in terms of the emitted power, range, and target properties. Nonetheless, the intensity values do not follow this
prescribed behaviour. Therefore, data driven approaches are used, allowing a better prediction of the observed intensity from the
range and reflectivity of the targets. For a Riegl LMS-Z420i and an Optech ILRIS 3D these experiments were performed. Both
scanners measure range by the travel time of a pulse. In our experiments, the reflectivity can be estimated from the laser scanning
data with a standard deviation of 6% or better. This demonstrates the potential for retrieving material properties of natural surfaces,
too.
1. INTRODUCTION
Obtaining geometrical information from terrestrial laser
scanning (TLS) is an established surveying procedure (Grün
and Kahmen, 2007, Fritsch, 2007) and used e.g. in cultural
heritage recording and industrial plant reconstruction. The
acquired point clouds, i.e. sets of xyz coordinates, are used to
determine object surfaces by triangulation, surface fitting, or
primitive instancing. Airborne laser scanning (ALS) is similar
with respect to the data provided: the point cloud. Calibration
of the ranging and scanning devices is an issue in the terrestrial
(Lichti 2007, Nothegger et al., 2007, Reshetyuk, 2006) and the
airborne case (Kager, 2006 and references therein). This allows
obtaining high precision, well beyond 1:10000. Laser ranging
uses energy emitted from the sensor for determining the range
between sensor and object. It is retrieved by measuring the two-
way travel time of the signal bounced back at the object.
Beyond the run-time it is possible to measure the strength of the
backscattered signal as well. Object properties like specular and
diffuse reflection behaviour, absorption, and transmission
influence the strength of this backscatter. The so-called
“intensity” value is related to the power (amplitude) or energy
of the returned signal. 1 With calibration it becomes possible to
convert these intensity values into parameters related to the
object surface. In ALS methods for radiometric calibration have
been proposed (Briese et al. 2008, Höfle et al., 2007 and
references therein). Independent thereof, these intensity values
have been used in TLS applications, e.g. for orientation (Akca,
2007), manual inspection of trees (Aschoff et al., 2004), and
rock face investigation (Rosser et al., 2007).
1 We use the term “intensity” in this paper, but there is not
necessarily a unique physical interpretation for these
intensity values by the different scanner producers.
In this paper we want to show that a radiometric calibration is
possible for terrestrial laser scanners as well. It builds on and
enhances previous work of our group (Pfeifer et al., 2007). The
next section gives motivations for doing this research. This is
followed by a section on the theoretical basis, discussing also
issues of not strictly monostatic laser rangers. Thereafter we
present our experiments, where a Riegl LMS-Z420i and an
Optech ILRIS 3D were used 2 . The calibration results and the
discussion follow in the subsequent chapters.
2. MOTIVATION
The overall aim is to extract more information than “only” the
xyz point cloud from TLS. This becomes possible if influences
on the loss of emitted energy in comparison to the detected
energy can be grouped into those depending on the object and
other influences, e.g. the distance from sensor to object. Not
only absorption and reflection properties, i.e. the BRDF, but
also the incidence angle of the measurement are counted in the
following to the object properties.
In many monitoring circumstances the objects observed are
known. One example is TLS for snow monitoring (Prokop,
2006) for research on and assessment of avalanche risk. In the
work of Prokop it is also observed that under certain
meteorological circumstances no range measurements are
possible, actually referring to an energy level too low to be
detected. Kaasalainen and Kukko (2007) advance this approach
with a more physical approach. In (Rees, 2006) it is described
how the grain size of snow and the snow temperature affect the
backscatter strength. From the intensity values it should
therefore, at least theoretically, be possible to reconstruct snow
2 Much of this paper applies to phase-shift systems as well.
1045