» Approximate 
image plane 
Point set 1 
   
   
  
Figure 6. Reflecting target perspective projection 
The Least Squares Matching (LMS) algorithm however requires 
a digital image with high contrasts between the searched point 
and the surrounding pixels of the object. 
This problem has been solved using a well known image 
enhancement technique: in the search image the reflecting target 
contrasts are emphasized using a digital filter (e.g. Wallis or 
High Boost). 
Figure 7 shows a practical example of the filtering procedure 
effects: after enhancement, the reflective target is well 
contrasted in comparisons to the rest of the image. 
Original search image Ennhanced search image 
   
   
  
Dolo Volue 
a 
3 
Dola Volue 
  
  
  
    
50 160 150 208 a 50 105 155 705 
Sample Sample 
Figure 7. Search matrix enhancement 
The set of the reflecting target coordinates extracted from the 
digital image that has to be oriented is named “point set 2” in 
the following. 
2.4 Homologous point definition 
The last step of the procedure connects each point of “point set 
1” to the homologous in “point set 2” to build the correct input 
data for the orientation parameter estimation. 
A ! 
  
  
c 
Figure 8. “Point set 1” and “Point set 2” 
In this case the software compares two planes: the [xy] plane 
defined in the previous paragraph where “point set 1” has been 
projected and the [En] of the digital image that has to be 
oriented, where "point set 2" has been located (see fig. 8). 
The software considers the four points of each set which 
circumscribe the other points as homologous. The homographic 
transformation parameters are estimated using the double set of 
coordinates for each point. 
All the remaining points of the “Point set 1” are then projected 
on the plane of the digital image using these parameters. For 
each transformed point of “Point set 1”, the nearest point of 
“Point set 2” is considered as the homologous. 
A complete test of the described procedures is described in the 
following paragraph. 
4. AUTOMATIC ORIENTATION OF A DIGITAL IMAGE 
OF THE RAMSES II STATUE 
The test of the above described software has been performed 
using the data acquired during the survey of the statue of 
Ramses II (catalogue #1380 of the Egyptian Museum — Turin, 
Italy). 
The aim of the test was to determine the orientation parameters 
of the digital image reproduced in figure 3. The image has been 
acquired using a Rollei 6008 semimetric camera provided by a 
calibration certificate. 
The first step of the procedure provided the automatic 
identification of the 14 reflecting targets placed on the statue 
(see fig. 2). 
Table 10 shows the estimated coordinates in the laser internal 
reference system. 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Target X Y Z 
# Im] [m] [m] 
20 -3,546 -0,149 -0,771 
21 -3,540 0,270 -0,832 
22 -4,118 0,323 -0,094 
24 -4,333 0,218 0,819 
25 -4,296 0,104 1,057 
60 -3,550 -0,146 -0,682 
61 -3,537 0,305 -0,665 
62 -3,742 0,325 -0,440 
63 -3,853 0,083 -0,074 
65 -4,265 0,099 0,113 
66 -4,274 -0,200 0,239 
67 -4,213 0,098 0,412 
68 -4,395 0,318 0,553 
Table 10 
The image shown in figure 2 has been taken with the camera 
near the tripod used during laser scanner acquisition; the 
coordinate of the perspective centre C have been assumed as Xc 
—- 03 m, Yc - 0 m and Zc - -20 m. The o, o and x angles 
have been considered null. 
The coordinates of table 10 have been projected by means of 
collinearity equation; the obtained coordinates x and y are listed 
in table 11. 
The second step performed the matching of the reflecting 
targets in the digital image. As it has been described before, the 
software uses the coordinates of [xy] plane as centres of the 
search matrices. 
Table 12 lists the coordinates (expressed in the fiducial 
reference system [&n]) of the reflecting targets founded using 
Least Squares Matching (as described in par. 2.3). 
—324-