Martin Smith 
  
Curve 1 (photo point data) = 375 points (average 8.3m between scaled points) 
Curve 2 (ground data) = 1180 points (average 2.7m between points) 
Number of matched points = 110 points 
RMSE plan = 1.1m 
RMSE height = 1.1m 
  
View along x-axis 
20 | 
45% 
eon 
Height 
OF 
  
    
  
400 2200 0 200 
| Northings ; 
Figure 7. View along the X axis showing the final position of the image curve and GPS curve 
(note: exaggerated Z scale; GPS curve = --- , model points = +++; all values in metres). 
  
  
  
Two options were considered, the first was to introduce a height into the 2D image coordinates by taking the height from 
the previous iterations matched ground point. Using the Campus photography the following results were obtained: 
Curve 1 (LH photo point data) = 1768 points (average 3.1m between scaled points) 
Curve 2 (ground data) = 8184 points (average 0.8m between points) 
Number of matched points = 918 points 
RMSE plan = 0.5m 
RMSE height = 0.0m (expected because of the method) 
Curve 1 (RH photo point data) = 1768 points (average 3.1m between scaled points) 
Curve 2 (ground data) = 8184 points (average 0.8m between points) 
Number of matched points = 1061 points 
RMSE plan = 0.5m 
RMSE height = 0.0m (expected because of the method) 
The second option is to use the curve matching to find corresponding photograph points to perform a relative orientation 
between the pair of photograph. The model coordinate curve is then used in a 3D-3D transformation to perform 
absolute orientation. The following results are from a matching of left and right hand Nottingham photo curves (1021 
points). The matched points are used to create a 3D model coordinate curve (relative orientation) which is then matched 
with the ground GPS curve (absolute orientation). The photograph curve coordinates were observed stereoscopically 
and therefore should produce a good match and result. The RMSE in plan = 0.8m (0.5m) and in height was 1.0m (0.5m) 
bracket values are those obtained from the absolute orientation results, see section 5.1. 
’ 
6 CONCLUSIONS 
The new technique to perform absolute and exterior orientation has produced some interesting results that have proved 
the viability of the method. It can be used in analytical and digital techniques and, if combined with an automatic linear 
feature extraction technique, has the potential for almost complete automation. The method has shown how ground, 
“real world” coordinates are captured in an easy manner using RTK GPS, although a wide variety of sources of ground 
coordinate data could be used (GIS, Digital Map, etc.). The curve data needs to be as dense as possible and have a wide 
  
856 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 
  
CC 
€} 
re 
ry 
re