In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
332 
620462.35 
4847283.10 
196.90 
8820 
3673 
8814.52 
3673.92 
8819.71 
3671.72 
620454.97 
4847305.51 
196.97 
8644 
3731 
8638.47 
3731.46 
8643.48 
3729.39 
Building 9 
620583.17 
4848170.14 
174.48 
1906 
2726 
1913.42 
2725.02 
1904.89 
2722.50 
620606.54 
4848099.71 
174.47 
2452 
2543 
2457.80 
2545.60 
2450.69 
2542.62~~ 
620633.01 
4848108.29 
174.30 
2385 
2338 
2392.24 
2340.99 
2384.92 
2337.57 
620609.43 
4848178.84 
174.35 
1840 
2519 
1846.87 
2522.00 
1838.13 
2519.04 
Building 10 
620084.29 
4847745.74 
178.76 
5183 
6598 
5182.56 
6594.73 
5182.64 
6599.30" 
620102.43 
4847691.31 
178.71 
5605 
6461 
5604.51 
6455.31 
5605.40 
6459.34 
620202.51 
4847724.94 
178.53 
5346 
5683 
5345.80 
5679.46 
5345.98 
5682.09 
620184.17 
4847779.22 
178.67 
4925 
5821 
4924.96 
5820.60 
4924.32 
5823.72 
Compared to image coordinates extracted manually 
Average(pixel) 
0.42 
-0.13 
-0.45 
-0.59 
RMSE(pixel) 
4.99 
2.66 
0.66 
1.49 
4. CONCLUSIONS 
In this study, we present a new method for registering existing 
3D building models with image data. Optimal building models 
are extracted with a priority function using information of 3D 
building model. Straight lines in the image are also extracted by 
the Bums algorithm. Optimal building primitives are projected 
into image space to compare both sets of data. Corresponding 
coordinate pairs are computed by similarity measurement, 
scoring straight lines contained in the buffer zone of the optimal 
building model. Finally, computed coordinates pairs are used to 
adjust the initial EO parameters. The proposed method for 
registering 3D building models with image data has been tested. 
The experiment showed that with optimal building models 
average differences of 0.43 pixel and RMSE of 1.62 pixel in the 
X direction and of average difference of 0.23 pixel and RMSE 
of 1.43 pixel in the Y direction were obtained. For the check 
building models the results were 0.43 pixel with RMSE of 0.66 
pixel in the X direction and 0.59 pixel with RMSE of 1.49 pixel 
in the Y direction. These results indicate that our proposed data- 
driven method can effectively register and align existing 3D 
building models with new acquired image data. Further work is 
needed to improve the proposed method by considering and 
including the errors of 3D building models and estimate their 
impact in the registration process. 
REFERENCES 
Bums, J. B., Hanson, A. R. and Riseman, E. M., 1986. 
Extracting Straight Lines. IEEE Trans. Pattern Analysis and 
Machine Intelligence, Vol. 8, pp. 425-445. 
Fonseca L, M. G. and Manjunath B. S., 1996. Registration 
Techniques for Multisensor Remotely Sensed Imagery. 
Photogrammetric Engineering & Remote Sensing, Vol. 62, No. 
9, pp. 1049-1056. 
Habib, A., Ghanma, G., Morgan, M., and Al-Ruzouq, R., 2005. 
Photogrammetric and Lidar Data Registration Using Linear 
Features. Photogrammetric Engineering & Remote Sensing, Vol. 
71, No. 6, pp. 669-707. 
Lumia, R., Shapiro. L. and Zuniga, O., 1983. A New Connected 
Components Algorithm for Virtual Memory Computers. 
Computer vision, Graphics, and Image Processing, Vol. 22, pp. 
287-300. 
Zitova, B. and Flusser, J., 2003. Image registration method: a 
survey. Image and Vision Computing, Vol. 21, pp. 977-1000. 
ACKNOWLEDGEMENT 
This research is supported by a grant (07KLSGC03) from the 
Cutting-edge Urban Development - Korean Land Spatialization 
Research Project funded by Ministry of Land, transport and 
Maritime Affairs of Korean government.