173 
CULTURAL HERITAGE WEB SITES: THE SUPPORT OF GEOMATICS 
Luigi Colombo 
University of Pisa, DIEIT - Sezione Topografia e Fotogrammetria, Pisa, Italy 
Barbara Marana 
Polytechnic of Milan, DIIAR- Sezione Rilevamento, Milan, Italy 
Commission VI - Working Group 3 
KEY WORDS: web site construction, Cultural Heritage inventory, photogrammetric imaging, photo-texturing. 
ABSTRACT 
In the field of Cultural Heritage inventory, geometric and thematic documentation assumes a very meaningful role to preserve the 
historical witness of real objects, as time goes by. 
The more current survey techniques (geomatics) take advantage of computer science, in order to manage procedures of 
representation connected to imaging, photo-texturing and 3D reconstruction. 
So, it is possible to perform thematic analyses (decay, materials, etc.) on the geometric raster model. 
This leads to an integrated process of acquisition that can be carried out in an exhaustive way inside a data transmission system like 
Internet, the worldwide network, and/or Intranet, the local subnet based on the same technology. 
The experience here described deals with some significant technologies to document and archive monumental sites inside Internet; 
besides, it presents the web site realised for the Harlequin village, an ancient little settlement close to Bergamo (northern Italy). 
1. GEOMATICS FOR THE CREATION OF VIRTUAL 
WORLDS 
The survey techniques, such as photogrammetry for architecture 
supported by computer graphics, are in fast evolution (Colombo 
et al., 1999) towards the processing of raster data (bitmap, 
imaging) in addition to vectorial information, inside a 3D 
reconstructed model. This procedure of visualisation transposes 
photos in raster format, by means of a scanning procedure, 
corrects perspective deformation, then scales and refers all in a 
spatial system (georeferencing). The management is performed 
through the tools of geomatics, which is substantially a 
discipline of engineering related to the collection and semi 
automatic processing of land data, with the aid of computer 
science. 
This way, a set of information (dimension, colour and spatial 
location) is assigned to all the pixel units of an orthogonalised 
image: the process gives to raster description the functionality 
of a numerical representation. 
These new working perspectives, thanks to recent improvement 
in computer technology, have introduced graphic space 
visualisations and survey methods more and more oriented 
towards automation and reality. 
So, it is possible to speak about imaging, virtual modelling and 
photo-rendering. 
The descriptive syntheses of the objects and the following 
thematisation, realised by manual and/or semi-automatic edge 
detection techniques on the image, seem to answer positively 
the growing worldwide demand for documentation and 
archiving of historical sites and building modelling. 
The digital image is used both indirectly for vectorial geometric 
plotting and directly for visual raster representation, such as 
surface texturing. The introduction of photo-textures improves 
the perception of constructive elements. 
However, there are still problems to enhance the 
correspondence between texture and reconstructed geometry: 
this is due to hidden parts of the object (occlusions), to shadows 
and generally to a lack of knowledge (for instance the shape at 
the ground level, textures of the façades, the structure of the 
roof, etc.). 
The surface reconstruction and the subsequent representation 
accuracy are therefore related to the morphology and to the 
requested degree of detail. Today, it's possible to collect in a 
semi-automatic way the spatial location of points on the 
examined object (image matching), to create the triangular 
meshes of a polyhedral surface approximating its shape and 
then to produce a shaded rendering and/or more advanced 
forms of photo-realistic visualisations. 
The knowledge through orthogonalised raster photos can be 
improved according to procedures of "multiresolution": the 
image or its portions are archived at different resolution levels, 
to let the user zoom in and out, without loss of definition, in a 
predefined range. 
According to different representation techniques, classic 
"vectorial data" use entity edges to describe geometric and 
thematic aspects or "isolines" to detail, as well as possible, the 
object morphology and spatial perception; on the contrary, 
"bitmap data" refer, for instance, to 2D orthogonalised and 
georeferenced raster images. 
These techniques allow both superimpositions of height grids 
and thematic layers. 
"Texture data" are oriented to render 3D Cad model of the 
object. The procedure gives realistic effectiveness to spatial 
descriptions: it is possible to speak about "virtual worlds", the 
computer product obtained through a wire-framed model of the 
object and the following superimposition of photo-bitmaps as a 
texture mapping. 
The quality of the geometric model, as accurate as requested by 
a selected degree of detail, and the congruence of the textures 
are prerequisites for good representations. These aspects are 
essential so that spatial visualisation can supply metrical values 
well correlated to the plotting scale.