You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Close-range imaging, long-range vision

and new
ki, Greece
> Subpixel
blem. In:
en (Eds.),
CV 2000
, Belgium.

Hiroshi TAKEDA Naoko KURISAKI Wentao CHE
Dept. of Spatial Information, Kokusai-Kogyo Co., LTD.
3-6-1 Asahigaoka, hino-city, Tokyo, 191-0065 JAPAN
{htake, naoko_kurisaki, wentao_che} @kkc.co.jp
Commission V, WG V/2
KEY WORDS: Airborne Laser Scanner, LIDAR, Digital Surface Model (DSM), Digital Elevation Model (DEM), Ortho Image,
Extraction of building footprints, Canny Filter, Hough Transform, Digital Maps (DM), 3D City Models
The Airborne Laser Scanner can collect the object surface point clouds efficiently for wide area within a short period of time. And it
has been recognized as available surveying tools in the field of 3D measurement. At the same time to the measurement of laser
scanning, we took the Aero Photographs by using the surveying digital camera. The purpose of this study is to automatically extract
the building footprints as the base maps in the urban area from the laser point clouds and ortho image. So far, the methods of
building extraction by using the laser point clouds or the aero photographs have been proposed. Referring to their studies, it is hard
for the extraction of building footprints from the Airborne Laser Scanner data. Because, in the case of measurement with the
Airborne Laser Scanner, the laser beams are not always guaranteed to hit the building edges, because the laser beams are irradiated
as a random direction. In the case of only using the laser point clouds, it is too expensive to obtain the evident building footprints,
because it is necessary to measure the small area with high density. In the case of only using aero photograph, it is difficult for the
extraction building footprints as closed polygons. In the other side, there is the demand for rough foundation data of maps at a low
price to create the maps. This paper proposes the methods of extraction the rough building footprints and the connection between
vertices from the laser point clouds and correcting building's footprints by using the ortho image to create a rough foundation map at
a low price.
1. INTRODUCTION Using the methods, the building footprints are obtained as
polygons, and we can use the data of building footprints
Recently, the expectations are increasing for the precised 3D directory into the GIS systems. And then, we can create easily
City Models according to the improvement of GIS and CG the 3d City map from the building footprints and the height
technologies. Specially, the precision Digital Maps that are information of laser points.
reflected the latest status of real world is indispensable for the
Virtual Reality System. In the meantime, the building footprints 2. SPECIFICATION OF AIRBORNE LASER SCANNER
are changed frequently as time goes on by the building
reconstruction or re-planning of the streets in the real world. In
order to keep freshness of the Digital Maps in a short time, we
need to create the 3D City Models, and need to maintain and
modify the data sets at the same time. In the case of 3D Digital
Maps are provided as the base maps, the data sets are required
to provide with uniform quality for wide area as well.
So far, we have studied the filters for removing point clouds
hitting objects on the ground and extracting the ground point
clouds (DEM) from the surface point clouds (DSM) in the
urban area. The surface point clouds means the all measurement
surface of objects on the ground that include ground, buildings,
trees and the others. The filters mean the method to classify the
ground and the others. As a result, we can get the DEM.
And then, the height of objects is obtained by the calculation
that subtracts between DSM and DEM. The incipient object
outlines as closed polygons are created to conduct the contour
lines from the height of objects data. At first, the incipient
object outlines are classified to the buildings and the trees. And
then, we use the buildings cluster as the incipient buildings
footprints. These outlines are more complicated than the actual
building footprints, because the laser beams hit a building
randomly. The evident building footprints are obtained by the
adjustment from the incipient building footprints using the
ortho image.
The specifications of using equipment are shown in table 1-2.
The laser scanner and the digital camera are loaded on an
airplane, and the instruments are synchronized with each other
to survey.

Laser Type Near Infrared
(Wavelength 0.7-1.3 um)

Laser Standard Class4 Laser Product
Flight Altitude 600 — 3000[m] (for ground)
Scanning Width 300 — 2500[m]
Scanning Angle S[deg](28Hz) - 45[deg](14.5Hz)
(Scanning Frequency)
Pulse Rate Max.15 [KHz]
Return Number Max. 5 [Pulses]
Measurement Interval Im — 8m
(Cross Track)
Measurement Interval Min. 2.5m
(Along Track) (Due to Flight Speed)
Accuracy Horizontal: +30cm
Vertical: 15cm (RMSE)

Table 1 Specification of Laser Scanner

Image Size 4096 x 4096 [pixels]
Shutter Interval Min. 2.5 [sec]
Focus Length 50 or 90 [mm]
View Angle 45[deg] fixed (F=50mm)
30[deg] fixed (F=90mm)
Ground Resolution 15cm (Max)

Table 2 Specification of Digital Camera
ar 3