while on the 
easured. 
. to construct 
data by EPI 
oints. In the 
hod utilizing 
to match the 
id shape of a 
10to analysis 
| map 
isured points 
map. These 
ion, such as 
red points in 
peak of this 
boundary of 
1e histogram 
| Z-axis with 
  
sional 
am 
ie building 
the track of 
cular line is 
  
  
  
drawn from the building’s edge facing the street toward this 
track. The record of intersection of this line and trace is utilized 
to make the pattern of buildings. 
The next step of process is to match the histogram of measured 
points by EPI analysis with the pattern of the buildings made 
from a map, then judge the location of the building boundary in 
the peak of the histogram. 
We apply DP matching method to match 3D measured data by 
EPI analysis with 2D digital map. 
  
  
  
  
The building 
boundary pattern 
  
= The track of the vehicle 
The perpendicular line from a building boundary 
Figure 7: Building boundary pattern by digital map 
3.2 Solid shape construction 
  
Figure 8: Solid shape example of a building 
including height information 
We utilized the aerial photo analysis result which was made 
with a stereo method. Using an aerial photo analysis we gave 
height information to a model made a digital map like figure 8. 
So we can make solid shape of a building. We put the texture on 
the model. 
4. TEXTURE MAKING BY PROJECTION 
CONVERSION 
The motion images taken from a camera on a vehicle reflect the 
upper part of a building. It is thought that the texture of the 
upper part of a building can be is gained by projection 
conversion of an image. It is assumed the plane of an parallel 
picture to the direction to which a vehicle goes in the right-and- 
left both ends of image like figure 9. We can make an image of 
screen of side where upper texture of a building appears in. 
However, resolution of an image becomes low in a made image. 
So, we make the consecutive image with screen of the side by 
projection conversion with the consecutive image. Then we 
match the image and integrate them. 
    
   
; The image by which projection 
conversion was carried out 
  
The souse image 
A camera position 
  
  
  
Figure 9: Generation of the image by projection conversion 
A vehicle progress direction 
    
   
   
   
  
* 
1»Yn+1 »Z A 
Hsaaskeng 
pA 
> 
  
  
  
  
Z 
The n+1-th fi are n» n) 
zssssusinssum 
Y 
The n-ttrfrante 
  
  
"À cdmera position 
  
The souse image 
X 
Figure 10: Matching of a position of an image 
It is shown in a figure 12. A point on a building (X,Y, Z), asa 
point on an image of a n frame (x n, y n ,z n), the image which 
AZ moved, a point on an image of a n +1 frame eye (xn+1,y n 
+1,z n +1), I turn the focus distance into f. If we can pursue 
distance X from a camera pass to a building, they was matched 
by an expression (1). 
Because the target which texture information wants to acquire is 
only a building facing a road, we aline it by demanding X of 
distance of a building facing a road than technique of Section 
3.1 and compose the image which I converted a projection into. 
A tm m LN 
X X 
n n + 1 
  
  
. EXPERIMENTAL RESULTS 
Un 
In these experiment, we used 700 consecutive input images built 
from the video image that we took by a car running along 
downtown. The car equipped with the gyro sensor and distance 
sensor in order to record vibration and to obtain the moving 
distance. GPS was used to record the location of the vehicle. 
Obtained image sequence was normalized using distance sensor. 
Because of real environment, it is necessary to revise vibration 
for image sequence. As a result of measurement of gyro sensor, 
—569—