The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B3b. Beijing 2008
594
Contrarily, for the urban traffic monitoring, in order to ensure
the system efficiency and derive the traffic flow information, a
much broader area is needed to be covered by laser scanner
surveying and multiple instances of vehicle object have to be
recognized and located from there simultaneously. It requires
more advanced algorithms to separate 3D vehicles laser points
from complex clutter surroundings. Under this situation, some
operations used for pose estimation or geometric inference are
not crucial as semantic decision of whether a vehicle exists or
not (vehicle counting).
In this paper we will study the feasibility and characteristics of
using ALS data to analyze vehicle activity in urban areas. Since
urban areas usually characterize dense road networks,
vegetation occlusion and anomalies (e.g. irregular structures
like wire, pole or flowerbed), we try to find out the optimal
laser data acquisition configuration for traffic monitoring in
view of reliability and efficiency, and propose conceptual
design of approach for vehicle detection and motion indication.
In this work initial research efforts are made to explore the
capability of solely using state-of-art commercial airborne laser
scanner for the task. The general and boundary conditions of
traffic analysis based on ALS are to be examined and outlined.
The purposed concepts and algorithms methods will initially be
assessed empirically in terms of accuracy and recognition rate.
Different impact factors on the results should be studied.
Moreover, an improved completeness of vehicle detection can
be expected due to penetration of laser ray through tree
canopies. The modeling of object under volume scatters is an
important issue for the recognition task in the 3D laser data.
The goal is to diagnose to what extent vehicles under trees can
be hit and sampled by penetrating laser rays, and further be
recognized and reconstructed by computer operations, even if
human inspection also cannot.
This paper is structured as follows: first, the configurations of
laser data recording in view of urban traffic analysis are
discussed and the vehicle models for stationary and moving
ones are introduced; next, general approaches for detecting
vehicle from urban laser data tending to derive traffic flow
parameters are proposed and analyzed; and finally, the
conclusions are presented.
2. LASER DATA ACQUISITION FOR URBAN
TRAFFIC ANALYSIS
Usually, traffic monitoring using LiDAR, as mentioned here,
refers to the direct collection of 3D information from airborne
platform rather than from ground-based sensor. Deriving the
traffic flow parameters statistically demands a certain spatial
coverage of data acquisition. Currently, ALS systems show a
great variability and flexibility concerning data acquisition
strategies; we want to first compare and analyze different
scanning configurations and attempt to qualitatively evaluate
results on the traffic analysis depending on different factors.
Generally, traffic - related information are expected to be
extracted as add-ons of regular LiDAR mapping systems,
together with topography and city models, so that current laser
surveying systems could be adapted to the solution to traffic
monitoring at no extra efforts. However, in the long term, one
may also think of operational traffic monitoring systems based
on ALS.
Current ALS systems work almost solely in the pulse time-of-
flight measurement principle for ranging, detecting a
representative trigger signal for multiple echoes in real time
using analogue detectors (Pfeifer & Briese, 2007). The direct
objective of ALS is to reconstruct 3D geometric model of
sensed environment as accurate as possible. Various system
specifications and relations have been examined in order to
clarify the scanning process and related impact factors on the
range accuracy (Baltsavias, 1999). However, via taking a deep
look into them, some parameters are also considered as being
relevant and sensible for the traffic-related analysis using ALS
data, which are listed as follows:
1. View angle, namely the angle between the scan plane and the
horizontal level
2. Surface sampling capacity — Footprint size, which is affected
by laser beam divergence and flight height and Point
density, namely point spacing which can be decomposed
into along-track and across-track components
3. Field of view, namely swath width which is determined by
flight height and range of scan angle
4. Scan pattern and relation between flight path and vehicle
queue
5. Minimum detectable object/energy
Being different from freeway and other open areas, such as
rural areas, urban areas face a more complex situation
concerning the traffic analysis from ALS due to dense road
networks, numerous buildings and vegetations, anomaly
structures. Any adjustment of sensor configurations can easily
lead to change of data characteristics, which may be exploited
for specific applications.
2.1 View angle. Concerning the view angle of ALS, normally,
it amounts to 90 degree, perpendicular to flight line, forming
the most common scanning geometry: nadir-view; if not
perpendicular, it then refers to forward - or backward looking
ALS. In case of oblique view (Hebei & Stilla, 2007), a side of
vertical structures such as building façade is recorded whereas
another side would cast a big shadow causing loss of
information about surrounding objects (Fig. lc, d). The oblique
view of ALS can also lead to abnormal incidence angle of laser
ray interacting with the illuminated surface, which has been
proven to be adverse to laser backscattering mechanism. Most
incident laser energy is scattered away in this case, especially
for vehicle surfaces which are constituted of mental (Fig. la, b).
Moreover, the travel path of emitted laser ray becomes longer
due to inclination. It is crucial for detection of those vehicles
beneath the vegetation, because the penetration rate of the laser
ray decreases and we can receive even fewer laser pulses
backscattered from the vehicle surface. Overall, to avoid the
missing laser data and consider material properties related to
laser incidence angle, the scan geometry of nadir-view is
required.
2.2 Surface sampling capacity. The footprint size and the
point density seem to be two most relevant parameters among
system configurations. But they are determined by independent
factors which can be selected before flight. Nowadays, most
commercial systems can achieve the point density of about 1-10
pts/ m~ with a footprint diameter of up to 50cm increase per
1000m distance. According to experience it seems to get better
detection results when the laser point density increases.
Normally only the object model, which is represented by laser
point samples with certain level of detail, can be found and
recognized. Furthermore, laser footprints should not be
overlapped with each other in order to ensure that the captured
surface information carried by each laser echo are not mixed.
