:veloped by the 
These problems 
r relics and lack 
  
< of images 
ne lasers, high- 
lution amateur 
s investigation 
€ bulb function 
ow only outline 
of relics can be taken and stored in memory while a relic is 
moving perpendicular to the digital camera. Ortho image can 
be acquired by synthesizing line image and Depth (DEM) 
image also can be acquired by synthesizing the line image 
since the Z-coordinate was given as the changing values of the 
actuator. If the camera calibration was performed previously, 
the image coordinate for each line image can be transformed 
into 2D object coordinate since the Z-coordinate was given. 
Additionally, fitting a band pass filter, line images for relics 
can be taken without dark space. However, imaging for shadow 
Calibration Mark 
  
Control Moving Pitch 
Digital Camera 
(Slide Movement by Actuator) 
A. 
A . x 
  
    
Sub-Position 
— Front- 
Position . 
Controller 
(Camera + Camera Actuator) 
   
  
  
Controller 
(Actuator) 
areas occurring in hollow areas on complicated relics or faint 
areas at the top of columnar relics are still issues. 
In order to resolve these problems, calibration marks are fitted 
to the flame, and stereo image for relics are taken previously at 
the two camera positions where the front-position (front of the 
relics) and sub-position (slid the camera to side way). 
Calibration marks were taken with relics simultaneously at this 
time. The digital camera is slide on the actuator in this paper. 
Line Laser 
  
  
  
  
  
  
Actuator 
(One Axis) 
  
     
PC 
  
T ~~ 
  
Control Camera Shutter and Slide Movement 
Fig.2 Refined Ortho Imaging System 
Table 1 System Components 
  
Digital Camera 
Dimage7 (Minolta) CCD:5.2M(pixel) 
  
Line Laser SMF-501L-670 (M)-10-30 (MORITEX) Wave Length: 670nm 
  
Actuator (One Axis) 
IS-S-X-M-8-60-500(IAI) Minimum Moving Step: 0.1mm 
  
PC (CPU & Memory) 
  
  
CPU: Pentium4 1.7MHz, Memory: 1024MB( 1GB) 
  
  
3. INTERPOLATION METHOD 
Figure 3 shows flow of refined system, and Figure 4 shows 
interpolating process. In this system, stereo images for 
calibration marks and relics are taken simultaneously and 
previously. These images are used for interpolating, and in the 
refined system, human intervention is requested in calibration 
process, but other processes can be performed automatically. 
In order to interpolate omission parts, comparison with central 
and ortho image is performed in this system, but automatic 
identification of shadow area and faint area is impossible only 
comparison with the both image due to different image mode 
such as ortho and central projection. 
In order to resolve this point, central projection image is 
formed by using ortho projection image and DEM image, and 
called as simulated central projection image (simulated image) 
in this paper. 
Consequently, identification of shadow and faint area can be 
performed by overlapping with simulated central projection 
image and ortho projection image. Figure5 shows identified 
omission parts. 
—485-