In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
620590.78 
4847522.45 
189.85 
6934 
2662 
6931.69 
2666.24 
6934.84 
2661.68 
620604.89 
4847527.20 
188.93 
6896 
2553 
6894.12 
2556.62 
6897.14 
2551.67 
620577.84 
4847607.41 
189.07 
6270 
2763 
6267.26 
2766.40 
6270.27 
2761.45 
Building 7 
620309.61 
4847485.66 
177.48 
7205 
4856 
7200.85 
4853.73 
7205.98 
4855.24 
620315.47 
4847468.22 
177.51 
7346 
4808 
7336.08 
4808.71 
7342.80 
4809.70 
620327.47 
4847472.26 
177.55 
7309 
4716 
7305.08 
4715.75 
7307.08 
4715.75 
620333.49 
4847454.58 
177.49 
7446 
4668 
7442.09 
4669.44 
7444.98 
4669.81 
620374.30 
4847468.44 
177.52 
7342 
4352 
7335.62 
4353.18 
7338.15 
4352.52 
620362.97 
4847503.61 
177.54 
7067 
4439 
7062.97 
4440.27 
7067.20 
4439.07 
Compared to image coordinates extracted manually 
Average (pixel) 
1.07 
0.52 
-0.43 
-0.23 
RMSE(pixel) 
4.55 
2.62 
1.62 
1.43 
3.5 Adjustment of EO parameters 
With the 30 pairs of building model primitives and their image 
coordinate extracted by similarity measurement, the EO 
parameters are adjusted in a least squares solutions as shown in 
table 2. 
Table 2. Adjusted EO parameters 
Initial EO parameter 
Adjusted EO parameter 
X(m) 
620455.282 
620454.647 
Y(m) 
4847674.264 
4847676.303 
Z(m) 
1632.24 
1629.578 
Omega(deg) 
0.06556 
-0.01266 
Phi(deg) 
0.14135 
0.12273 
Kappa(deg) 
-89.86594 
-89.88080 
3.6 Assessment of the proposed method 
In order to evaluate the performance of the proposed method, 
qualitative and quantitative assessments were carried out with 
both optimal and check building models, respectively. Both 
optimal and check building models are back-projected into 
image space using the adjusted EO parameters. Figure 13 shows 
the results of each optimal building projected into image space 
after the EO parameters are adjusted, and Figure 14 shows 
better matching of the building edges using the new EO 
parameters than the initial EO parameters when compared to 
Figure 12. Reference coordinates corresponding to optimal and 
check building models were also manually extracted from the 
image for quantitative assessments (Table 1, column (1) and 
Table 2, column (1)). The reference coordinates are compared 
to coordinates of optimal building automatically extracted in 
Table 3. While the results with initial EO parameters show that 
the average difference in X and Y directions are 1.07 and 0.52 
pixels 
respectively, with RMSE of 4.55 and 2.66 pixels respectively, 
the results with the new EO parameters show that the average 
differences in X and Y directions are 0.43 and 0.23 pixels, with 
RMSE of 1.62 pixel and 1.43 pixels, respectively. The test is 
also carried out in a similar manner with check building models 
which were not used in the registration process. The average 
coordinate differences of check building models with initial EO 
parameters were 0.42 and 0.13 pixels in X and Y directions, 
with RMSE of 4.99 and 2.66 pixels, respectively. After 
adjusting the EO parameters, the result show that the average 
differences in X and Y direction are 0.45 and 0.59 pixels, with 
RMSE of 0.66 and 1.49 pixels, respectively. 
Figure 13. Back-projection results after adjusting EO 
parameters 
Figure 14. Enlarged image of the back-projected Building 4 
using the new EO parameters 
Table 3. Quantitative assessment with check buildings 
Check building vectors 
Image coordinates of 
check buildings extracted 
manually(l) 
Image coordinates of 
check buildings obtained 
by initial EO parameters 
Image coordinates of 
check buildings obtained 
by automatically adjusted 
EO parameters 
X(m) 
Y(m) 
Z(m) 
X(pixel) 
Y (pixel) 
X(pixel) 
Y(pixel) 
X(pixel) 
Y (pixel) 
Building 8 
620426.84 
4847296.02 
196.82 
8717 
3951 
8712.04 
3952.48 
8717.21 
3950.73 
620434.24 
4847273.58 
197.00 
8894 
3894 
8888.84 
3894.82 
8894.19 
3892.93