- 106 - 
scanning parameters (occlusion detection, accuracy, time 
optimisation...) 
All these reasons underline how is important to use digital 
instruments during the survey that allow to verify on the field, 
in real time, the data transferred on a portable PC and 
evaluating the best position of the point of view for the next 
scan or intensity image acquisition. 
Figure 1. Scan field of view on sacresty plan. 
The distance, between the center of the sensor and the points on 
the surfaces, has remains between 10 and 15 meters, so 
considering that the laser mesh was composed from 1000x1000 
points with an aperture of 40°x40°, the distance between the 
points, in the cloud that describes the 3D model, result included 
between 1cm and 1,5cm. The whole process of evaluation to 
create a 3D model took about 7 days, fairly divided between 
geometrical mesh construction and textures application. 
A reconstruction algorithm based on polygonal meshes was 
used to produce a triangulated model. 
Figure 2. Triangulated model of a scan. 
To map the texture upon the surface covered from a single scan, 
has been necessary two photographic images taked from a 
position close to the laser scanner viewpoint. The digital camera 
adopted is CANON Powershot PRO 90 IS with CCD 3,34 
Mpixel sensor. 
One possible solution for determining the relationship between 
range and color images is through calibration using the 
calibration board and fixtures. However, this method requires 
that the range and color sensors be fixed on the fixture once the 
relationship is calibrated. But usually a color camera is much 
handier than a range sensor so is better to take color images 
freely without having to transport a heavy and fragile range 
sensor. 
As side product of range images, range sensor often provide 
reflectance images that represent a collection of the strenght of 
returned laser energy at each pixel. This image is aligned with 
the range image because both images are obtained through the 
Figure 3. Reflectance image of a scan. 
same receiving optical device, in other words reflectance and 
range data are fully registered, considering they both originate 
with the same echoed laser. The returned timing provides a 
depth measurement, while the returned strength provides a 
reflectance measurement. A reflectance image is itself an image 
of the scene and can be matched to any other image such as 
photographic images. 
Our range sensor Cyrax provide reflectance image, so we 
decided to employ this image as a vehicle for the alignment of 
range images with intensity images. 
Reflectance images are similar to intensity images in that both 
images are somehow related by surface roughness. Since the 
reflectance image is aligned with the range image, so to align 
the reflectance image with the intensity image is much easer 
task than that aligning the range image with the intensity image. 
Corrispondences between the reflectance and the digital images 
are computed using a semi-automatic system, that align 
extracted features (edges or points) from reflectance images 
with those in intensity images. It is possible to guide the system 
by adding extra points and find further corrispondences. These 
edges are easy to find in the intensity images, since they are 
generates from a boundary between two different colors or 
materials that generates a discontinuity of reflectance. Thus, this 
method allows to align 3D points on range surfaces with 2D 
points in the intensity images. This is very flexible in that it 
allows the model to be texture-mapped using an image taken 
with any camera, at any time (even historic images can be 
uesed), from any location. The software used to process the 
laser scanner data is RECONSTRUCTOR realized by 3Dveritas.