-22- 
merged data set is shown in Figure 8. The side view clarifies 
that the undercuts were well digitized. 
Figure 9. 3D-Projection of Diplodocus camegii 
A very challenging device was a dinosaur skeleton of a 
Diplodocus camegii at the Museum of Nature Humboldt - 
University of Berlin. The skeleton had a length of 25 m and was 
5 m high. 21 laser scanner images were taken. These 
independent data sets were transformed into one common 
coordinate system by adjustment procedures. Figure 9 shows 
the final result in 3D-projection. The skeleton is imaged by 
861021 laser measurement points. 
4. PROCESSING OF LASER SCANNER DATA 
The presented examples make clear that laser scanners generate 
point clouds with a large amount of data which cannot be 
processed by standard CAD programs. This chapter addresses 
the problems of processing the data so that they can be input 
into CAD software packages and of generating NC-programs 
for re-production in archaeology. Solutions for merging 
effectively independent laser scanner images into one common 
coordinate system will be presented. 
4.1 Procedures for Merging Laser Scanner Images 
The previous chapters clarified that several independent laser 
scanner images must be merged when the object is larger than 
the field of view of the laser scanner or the object exhibits 
undercuts. A very straightforward solution is using identical 
points in the overlapping area of neighboring images. This 
requires that these points appear with high contrast. In cases 
where identical points cannot be identified on the object itself 
so called reference spheres can be applied (Wehr, 2001). Using 
these aids very precise results can be yielded. 
The Cartesian coordinates that are the result of the digitizing of 
freeform objects by the 4D-LM are transformed into object 
coordinates by a special software postprocessor. The 
coordinates may now be processed as either NC-commands for 
copymilling (duplicating milling) and rapid prototyping 
machines or in the data format for the sculptured surface 
modeller. 
4.2 Postprocessing of Object Coordinates for Copymilling 
and Rapid Prototyping Machines 
A special software functionality is developed for the calculation 
and determination of loop contours out of the digitised data. 
Figure 10 and 11 illustrate the method for the estimation of the 
contours for the front digitised view of the glass head in figure 
6. In further steps of research while analyzing NC-programs of 
copymilling machines, we found out that the coordinates of the 
cutter center point represent a set of loop contours. This means 
that the calculated contours could be transformed directly to NC 
Programs. 
This method presents a simple way of how even non experts in 
CAD/CAM systems could generate and produce a 3D-modell. 
Figure 10. Estimation of the contours 
A more advanced method is to carry out the merging after 
surface modelling. 
The tooling and moulding industry currently models the work 
pieces mathematically through Computer Aided 
Design/Computer Aided Manufacturing- (CAD/CAM) systems. 
Unfortunately the analytical description of such objects is not 
sufficient for practical applications. Here the computer space 
and time complexity must be increased to achieve the 
approximation and interpolation necessary. 
Digitizing a sculptured surface results in a very high amount of 
Figure 11. Estimation of the contours 
data (Giga Bytes of space in the main memory system). The 
computation performance of existing CAD/CAM-systems is 
insufficient for the processing of this data. 
A data reduction is required to solve this problem. A data 
reduction of more than 80% can be achieved depending on the 
complexity of the workpiece by computing spline -curves and 
-surfaces using the following algorithms 
bicubic Bezier, 
polynomial representation (Coons), 
B-Spline,