2004 
  
  
EFFICIENT LINE MATCHING BY IMAGE SEQUENTIAL ANALYSIS 
FOR URBAN AREA MODELLING 
Y. Kunii*, H. Chikatsu 
Tokyo Denki University, Department of Civil and Environmental Engineering, 
Hatoyama, Saitama, 350-0394, Japan - (kunii, chikatsu)@g.dendai.ac.jp 
Commission Youth Forum 
KEY WORDS: Urban, Extraction, Sequences, Matching, Modelling, Visualization 
ABSTRACT: 
3D city modelling from airborne imagery includes mainly two parts: (1) image processing procedures and (2) 3D modelling for man- 
made objects such as buildings, roads and other objects. Line extraction and sterco matching are usually utilized as an image 
processing procedures. However, there are some issues for automatic man-made object modelling. In particular, spatial data 
acquisition of buildings are important for reliable city modelling 
With this objective, this paper focuses especially on efficient line matching method using optical flow and trifocal tensor. 
Furthermore, line matching in general stereo matching methods were also investigated, and performance of the proposal line 
matching method was compared with these general methods in this paper. 
1. INTRODUCTION 
Recently, efficient spatial data acquisition and visualization 
have been receiving more attention from the view point of city 
planning, city regeneration, telecommunications, environmental 
and energy problems. Generally, in order to perform object 
modelling using digital image, line or feature extraction and 
stereo matching are performed, and many matching methods 
such as area based matching, future based matching have been 
proposed. In particular, line gives important information for 
building extraction, and satisfied 3D results are depend on 
rigorous line extraction and matching. 
With this motive, the authors have been concentrating on 
developing an efficient line matching procedure for man-made 
object modelling using high vision imagery (Kunii and 
Chikatsu, 2003). The line matching is comprised of line 
extraction and line tracking, line extraction was performed by 
Canny operator (Canny, 1986), and line matching was 
performed using optical flow estimation and epipolar matching. 
However, more efficient line matching is needed for automatic 
3D modelling due to fragment or multiple lines during the line 
matching procedure. 
In these circumstances, this paper focuses on more efficient line 
matching method using trifocal tensor (Beardsley, et al., 1996). 
The trifocal tensor is geometric relation of 3 images, and useful 
for point or line matching of multiple image. Therefore, the line 
matching for the high vision imagery could be performed more 
efficiently. 
Furthermore, line matching by general stereo matching methods 
were also investigated in this paper, and performance of the 
proposal line matching method was compared with these 
general methods. 
Finally, 3D modelling for the urban area was performed by the 
line information in this paper. 
  
221 
2. HIGH VISION IMAGERY 
The high vision imagery in this paper was taken by high 
definition television (HDTV) format. This paper reports the 3D 
modelling method using the high vision imagery which 
obtained from a helicopter at urban district of Kobe-city, Hyogo, 
Japan. Table | shows the major components of the high vision 
imagery, and Figure 1 shows the first frame imagery. 
Table 1. Major components of the high vision’ imagery 
  
LEN D ©0006 © 0 0 2 0 0 0 0 0 0 
  
® 000 000000 ©0000 ©0000 000000 00000000000000000000000000000 
  
  
  
  
  
,J 
3 
ey iy A end I 
ag d “ih 
  
Figure 1. First frame imagery 
3. LINE MATCHING BY OPTICAL FLOW 
The line matching was performed by line extraction and line 
tracking using optical flow. Detail procedures of the line 
matching method are as follows.