International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
     
LIDAR 
(Point cloud) 
Aerial Image 
  
3D Planar Patch 
Forming 
Straight Line 
Extraction | 
d 
SMS Method 
| 
Building Model | 
Figure 2. Flow chart of building reconstruction. 
4.1 3D Planar Patch Forming 
The first stage of building reconstruction is to extract the 3D 
planar patches from LIDAR data. A TIN-based region growing 
procedure is presented for 3D planes forming. The coplanarity 
and adjacency between triangles are considered for TIN-based 
region growing. Two coplanarity conditions are considered for 
merging triangles: (1) the angle between normal vectors for 
neighboring triangles, and (2) the height difference between 
triangles. When the triangles meet the coplanarity criteria, the 
triangles will merge together as a same plane. Once the planar 
segments are extracted, we use least squares regression to 
determine the coplanarity function of planar segment. Figure 3 
is the illustration of formed 3D planar patches. 
  
(b) 
Figure 3. Illustration of 3D planar patching. 
(a) triangular form in building region. 
(b) extracted 3D planar facets. 
42 Initial Building Edge Detection 
After extracted the 3D planes, we detect the initial building 
edge from raster form LIDAR data. The initial edges in single 
building region are obtained by a Canny operator (Canny,1986). 
A 2D line tracking is applied to the edge detection result. Then, 
we select a length threshold to remove short lines. After 
removing the short lines, we consider the elevation information 
of edge to perform the 3D line tracking. Each line is classified 
to the 3D planar patch as stated. The steps of initial building 
edge extraction are illustrated in Figure 4. 
  
   
@ 
Figure 4. Illustration of initial building edge detection. 
(a) DSM in building region. 
(b) edges detected from Canny operator. 
(c) filtering the short edge by 2D line tracking. 
(d) classified the edges to 3D plane by 3D line tracking. 
4.3 Straight Line Extraction 
Based on the rough edges, the precise building edges are to be 
extracted in image space. Through the Hough transform 
(Hough, 1962), straight lines in image space are detected. Given 
the image coordinates and the height information from 3D 
planes, we can calculate the 3D edges in the object space by 
employing exterior orientation parameters. Figure 5 is an 
example of straight line extraction. 
  
(a) (b) 
Figure 5. Building boundaries extraction. 
(a) 2D building lines 
(b) 3D building lines 
4.4 SMS Method 
After the straight line extraction, the accurate 3D edges are 
combined with the already developed Split-Merge-Shape, SMS 
method (Rau and Chen, 2003) for building reconstruction. The 
Split and Merge steps are the two main procedures for topology 
reconstruction from non-related roof-edges. The Shape step 
uses the available roof-edge height information to define an 
appropriate rooftop. 
734 
  
Internal 
The LII 
Chu Sc 
was obt 
points 
previou 
pts/m°. 
satellite 
The rela 
In build 
LIDAR 
size of | 
Figure ^ 
are rect 
orthoim 
8 and F 
in Figur 
1/1000 
classific 
Figure 
topogra] 
extracte! 
and les 
building 
the clas: 
only co 
accuracy 
Acquire 
Date 
| Resoluti 
Data 
2741400 
2741300 
2741200 
2741100- 
2741000 
250400