Hiroshi Masaharu
2 METHOD
The method is extracting buildings from high-density elevation data obtained by a laser scanner carried on a helicopter
by applying region segmentation method to the elevation data. Basic idea is as follows. Each building has a distinct
height difference from the surroundings. Therefore we can distinguish each building by segmenting the DSM (digital
surface model) with the condition that a pixel having the height difference within a predetermined threshold from a
neighboring pixel belongs to the same region.
Boundary data of building polygons are obtained from the segmented image. By combining the polygon data obtained
by this method with the original DSM, three-dimensional scenery of urban area is generated. Outline of the method is
given here and more detailed explanation of the method is written in Masaharu et al.(1999).
2.1 Region Segmentation
The region segmentation is carried out as follows. The height differences between neighboring grid points are checked
against a predetermined threshold value. If the difference is within the threshold, then the grid point belongs to the
same region as the neighboring grid point and they are assigned the same label number. Four-neighbor relationship is
used to judge connectivity. The laser scanner data used here are given as elevation data at grid points with 0.5 m
regular grid interval representatively.
Before applying the region segmentation, 3 x 3 median filter was applied in order to eliminate isolated point noise in the
original data.
2.2 Boundary Tracking
Labeled image that has label numbers of segmented regions as pixel values is created as the result of region
segmentation. Then boundary tracking of each segmented region is carried out. Because the resulting coordinate data
of vertices of the boundaries are much redundant, unnecessary vertices were eliminated using Douglas-Peucker
algorithm (Douglas and Peucker, 1973). In this way, regions with distinctive height difference with their surroundings
are extracted in the form of polygon data.
2.3 3-D Model Generation
Next step is to combine these polygons with the height data of the laser scanner. A 3-D city model was generated by
applying a program originally made for generating 3-D city models from building polygons of a digital map and laser
scanner DSM (Kamiya and Hasegawa, 1999), to the polygon data obtained from laser scanner data.
3 RESULTS
3.1 Laser Scanner Sensor Used in the Study
The data were taken by a heliborne laser scanner system. The pulse frequency is 20000 pulses/s and the scanning
frequency is 25 lines/s. The scanning width is 60°. These figures give the ground resolution of about 0.5 m in both
cross track and along track direction when the helicopter flies 250m above the ground at a speed of 50 km/s. The beam
width is 2.5 mrad which corresponds to also about 0.5 m spot size on the ground. For more detail about the
specifications of the sensor system, please refer to Masaharu et al.(1999).
CCD linear array color optical sensor is attached near the laser scanner. Thus we can get optical images
simultaneously and digital ortho image can be created (Hasegawa et al. 1998; Murakami et al., 1999)). Although this
feature of the sensor system was not used in the 3-D model construction written here, possible applications of this
feature are discussed.
3.2 An Example of Minokamo City
Minokamo city is an intermediate scale city with about 46,000 population in Gifu prefecture, central Japan. The
central area of the city was selected as the test site. There are densely gathered traditional houses and rather small
number of middle to high-rise buildings.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 557