ACCURATE TEXTURE-MAPPED 3D MODELS FOR DOCUMENTATION, SURVEYING 
AND PRESENTATION PURPOSES 
Matteo Sgrenzaroli, Erik Wolfart 8 
a 3DVeritas, Via Zutti 1, 21018 Sesto Calende (VA), Italy 
{matteo.sgrenzaroli, erik.wolfart}@3dveritas.com 
KEY WORDS: 3D Reconstruction, Visualization, Laser scanning, Texture, Photo-realism, Architecture, Cultural Heritage 
ABSTRACT: 
We present a software system to create texture-mapped 3D computer models of existing large-scale objects such as building interiors 
or exteriors, monuments, bridges, caves, etc. As input it uses range data from different laser range scanners and 2D colour 
information from standard digital cameras. The programme creates a triangular 3D model from the range information and maps 2D 
information on the 3D model to create the final result. We have tested the software by scanning and modelling a number of different 
objects. This paper presents the lessons we have learned for a successful data acquisition and discusses a range of software modules, 
which we consider important for processing laser range data in the context of architectural and Cultural Heritage applications. 
1. INTRODUCTION 
The conservation, maintenance and presentation of Cultural 
Heritage and in particular historic buildings (built heritage) are 
issues with increasing interest to the public. It recently obtained 
a lot of political attention (in particular in Europe) and continues 
to attract public and private funds. 
Missing documentation is a major problem during conservation 
of historic buildings. A precise documentation of the status quo 
is essential for the protection of a building, for scientific studies, 
during restoration and refurbishment, but also for the 
presentation to the general public. The requirements for the data 
acquisition are much higher than for modem buildings since the 
surfaces usually are not planar and decoration and ornaments 
can be recorded only with high-resolution spatial data. Many of 
the objects (e.g. carved ceilings) can simply not be model in 3D 
with traditional technologies, as the required density of 
measurements that cannot be acquired within a reasonable time 
frame. 
The possibility to merge 2D information with the 3D model is 
particular important in the area of Cultural Heritage for two 
reasons: 
1. For a more complete documentation. Historic buildings 
often contain important colour information like frescos and 
mosaics. Therefore, documentation is not only restricted to 
capturing the 3D geometry of a building but also includes its 
colour. Furthermore, images (in the visual spectrum and even 
more in the non-visible spectrum) can transport important 
information about the structural condition, the pollution and the 
history of a building. 
2. Visualisation is becoming ever more important in Cultural 
Heritage. As Cultural Heritage institutions are usually spending 
taxpayer’s money, they face increasing pressure to justify and 
present their work. Additionally, many monuments are remote 
or closed to the public to avoid any damage. Using photo 
realistic 3D models for VR presentations over the Internet or on 
CD-ROMs is way to educate and increase the awareness of the 
public. 
Common criticisms about laser scanning are that 
• The point cloud acquired by a laser scanner is difficult to 
handle due to the huge amount of data and the unintuitive 
way the object is represented. 
• It is often faster to get to the required result (e.g. 2D tables, 
sections, drawings, orthophots) using traditional 
technologies since the processing of the scanner data 
involves a lot of interactive work. 
Although the later can be true for simple applications, we think 
that many of the problems evolve from errors during data 
acquisition phase and from the lack of suitable processing 
software. Therefore, this paper will give a brief guideline for a 
successful data acquisition and discusses a suit of algorithms, 
which are required to process the range and colour data within 
the context of architectural and Cultural Heritage applications. 
The paper is structured as follows. In section 2 we will briefly 
describe a number historical buildings or monuments, which we 
surveyed using different laser scanners and 2D images. Section 
3 lists a number of problems faced during data acquisition and 
gives a guideline on how to acquire the data in order to achieve 
the optimal final result. Section 4 discusses various software 
modules required to process the data and obtain a 3D texture- 
mapped model, section 5 presents some software tools, which 
allow to exploit the 3D model for architectural and Cultural 
Heritage applications and shows a number of results. 
Conclusions and future work are reported in sections 6 and 7. 
2. CASE STUDIES 
Over the last months we scanned and modelled several large- 
scale objects: historical buildings, monuments and 
archaeological sites. We present five examples, which were 
selected to be representative for different surveying problems. 
Table 1 classifies these examples according to i) objects 
surveyed from inside or outside (i.e. internal rooms or external 
façade), ii) type of laser scanner used. Although the sample 
objects are different in their nature (architectural characteristics, 
dimension, location, presence of painting, fresco or decorations) 
they were all surveyed with the same data acquisition and 
processing technology. The following list briefly describes each 
of the objects: 
• Sagrestia Archivio di Stato, Mantova (Italy): Room 
annex to SS. Trinità Church, which was built around 1600 
and is characterized by complex stucco decorations on the 
dome. Survey results will be used for both, during 
restoration and for visualisation purposes. 
• Ex Hotel Colombia Genova (Italy): Built between 1921 
and 1927, it is characterized by neo renaissance decoration 
in the entrance rooms. The survey was done in the context 
of the hotel restoration to be used as university library in 
Genova.