-16- 
a) 
B 
Figure 1. Digital photogrammetry of the mosaic: a) one of the 4 
digital pictures taken on site for photogrammetry. A dashed 
rectangle highlights one of the surfaces also acquired with a 
structured light range camera; b) results of photogrammetry. 
These plates were also acquired with both 3D scanning systems, 
keeping in the framed area both the targets and portions of 
interest of the mosaic that can be acquired in this way at high 
resolution. The dashed rectangle in figure la highlight one of 
the areas taken at high resolution with the Optonet range 
camera. 
In figure 2 a couple of hi-resolution models generated with the 
pattern projection system are shown. It is interesting to notice 
how texture (figure 2a) and geometry artificially shaded (which 
is equivalent to raking light) (figure 2b), may give a different 
level of interpretation of the same data. For example in figure 
2b it is much easier to identify the lacking tesserae of the 
mosaic. 
By identifying each barycenter of the optical targets in the 3D 
images, groups of 3D coordinates corresponding to the different 
targets, generated by the two sensors, can be obtained. Figure 
lb shows for example a the set of targets generated by 
photogrammetry. From these measurements transformation 
matrices have been calculated employing the well-known unit 
quaternion method (Horn, 1987). Each matrix allows for the 
roto-translation (pose) of the 3D images from the local 
coordinate system of the range camera to an accurate global 
coordinate system determined by the digital photogrammetric 
procedure. 
The TOF 3D scanned was useful for generating a dense cloud of 
points representing the underground wall and a structure located 
in the middle of the mosaic (see fig. la) whose meaning is still 
unknown. 
In figure 3a the final “multiscale” model is presented, where 
the hi-res 3D models and the TOF acquisition are both 
reoriented to the photogrammetry coordinate system through the 
method described. 
b) 
Figure 2. 3D acquisition of the mosaic section near to target C: 
a)model with texture; b) model without texture and synthetic 
shading.