ISPRS, Vol-34, Part 2W2, "Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
several different images. This may be caused, for example, by 
overlapping in the image mosaic generated by the camera, or 
images taken from radically different views. There is also the 
case where the camera has zoomed to get very high-resolution 
data for a particularly important scene feature. 
The availability of multiple texture sources for a single triangle 
necessitates additional consideration. To decide which texture 
source to use, we make use of the label map obtained during 
hidden texture removal to provide the number of pixels in the 
corresponding image that would be used to texture map any 
given model triangle. All that is required is to sum the number 
of instances of the triangle in the label map. This is done for 
each triangle in every image and the image contributing the 
largest number of pixels is used as the texture source. Such 
that all those triangles sharing a common texture source are 
grouped together. 
Figure9 Texture mapping Result of Kyoto Train way Station 
area 
This work was finished by using Softimage|3D system. We 
summary the main procedure is shown as follows(The result is 
shown in Figure9): 
• Surface segmentation for 2D texture mapping 
• Generation of texture images by 2D affined 
transformation 
• Imputing texture images into Softimage|3D (TIFF to 
PIC) 
• Object oriented texture mapping based on 
Softimage|3D 
• Defining parameters for animation (color, camera, 
light, material and so on) 
• Visualization 3D virtual environments based on 
Softimage|3D 
• Output 3D results from Softimage|3D to TIFF images 
3.6 Spatial Object Editing 
After generation of 3D visual models or VR environments, we 
should check the result based on exiting air photos of maps. 
The task was finished base on the 3D editing environment of 
Softimage|3D system. We should make the procedure for 
sequentially highlighting 3D objects on a 2D base map and 
image environment. The checking are semi-automatically done 
by operator’s mouse processing. If we find the problem, we can 
lock this object and modify it based on different 3D editing 
tools. 
3.7 Problem and Future Application Plan 
According to our research result, we can find the following 
problems that should make feather research in the future: 
1) . The accuracy and density of laser range data 
• The distance between random points is too far. 
• Some site has no 3D data 
• The accuracy is lower for extraction buildings with 
high densities 
2) . Improving the accuracy of object structures 
• Improving the accuracy of object structures by using 
3D data from multi-sources (mobile mapping, laser 
range data from ground stations) 
• Automated 3D feature extraction based on 
knowledge bases and existing GIS 
3) . Improving the accuracy of texture images 
• Getting the orientation parameters by using digital 
photogrammetry system 
• Generation of ortho-photos for accurately texture 
mapping 
3.8 Application Plan 
Our future works can be summarized as follows: 
1) . Multi-sensor integration 
• Airborne laser range data processing 
• Ground laser range data processing 
• Mobile mapping data processing 
• Integration 
2) . Improvement of digital photogrammetry system for 
structure feature extraction 
• Automated extraction of roads and buildings based 
on DSM and laser range data 
• Automated extraction of roads and buildings based 
on existing GIS 
• 3D environment for editing 3D spatial objects 
3) . 3D GIS and virtual environment 
• Generation of 3D spatial information systems 
• Generation of virtual environments for visualization 
and simulation 
• Different applications 
4 References 
1. Besl, P.J., Jain, R.C., 1985. Three-dimensional object 
recognition. ACM-Computing Surveys 17 1,75-145. 
2. Besl, P.J., McKay, N.D., 1992. A method for registration 
of 3-D shapes. IEEE Trans. Pattern Analysis Machine 
Intelligence 14. 2, 239-256. 
3. Canny, J., 1986. A computational approach to edge 
detection. IEEE Trans. Pattern Analysis Machine 
Intelligence 8 6, 679-698. 
4. Castro, J., Santos, V., Ribeiro, M.I., 1998. Multiloop 
robust navigation architecture for mobile robots. Proc. 
IEEE Int.Conf. On Robotics and Automation, ICRA98, 
Leuven, Bel-gium, pp. 970-975. 
5. Chapman, D.P., Deacon, A.T., Hamid, A., 1994. 
Hazmap: a remote digital measurement system for 
work in hazardous environments. Photogrammetric 
Record 14 83, 747-758. 
6. Chen, X.Y., Medioni, G„ 1992. Object modelling by 
registration of multiple range images. Int. J. Image 
Vision Computing 10 3, 145-155. 
7. Brunn, A.; Guelch, E.; Lang, F.; Foerstner, W. 1998. 
Hybrid Concept for 3D Building Acquisition. ISPRS