CIPA 2005 XX International Symposium. 26 September - 01 October, 2005, Torino, Italy 
COMBINATION OF LASER SCANNER DATA AND SIMPLE PHOTOGRAMMETRIC 
PROCEDURES FOR SURFACE RECONSTRUCTION OF MONUMENTS 
Ch. Ioannidis (a> , N. Demir (b) , S. Soile ,a) , M. Tsakiri <a) 
(a) School of Surveying Engineering, National Technical University of Athens, Greece 
,b) Yildiz Technical University, Istanbul, Turkey 
KEY WORDS: 3D, Integration, Laser Scanner, Model, Reconstruction 
ABSTRACT 
The reconstruction of precise surfaces from unorganised point clouds is an important problem in terrestrial laser scanning 
applications. The goal of surface reconstruction algorithms is to approximate the correct geometry, topology and features of an 
unknown surface only through a finite set of sample points. But unless the input data satisfy certain properties required by the 
algorithms, such as good distribution, density and little noise, then the reconstruction program may produce incorrect or even fail to 
give results. This is a typical problem with cultural heritage monuments, mainly due to the user time involved in acquiring data from 
occluded parts of a monument of large usually size rather than the inability of the laser scanner to provide data. This paper proposes a 
solution to surface reconstruction that is incomplete by applying simple photogrammetric procedures, and specifically the application 
of bundle adjustment methods using monoscopic measurements of homologue points on multiple images. Thus, without the need of 
control points, with low cost software that does not require specialisation in photogrammetric knowledge and equipment, the surfaces 
of monuments can be fully reconstructed. A case study is presented in this paper using data collected by a Cyrax2500 laser scanner 
and images captured by a SONY DSC-F707 5Mpixel digital camera from a Byzantine church in Peloponnese, Greece. The use of the 
laser data only in a standard modelling software produced an incomplete surface. A commercial photogrammetric software was also 
employed to perform the self-calibration of the camera and the 3D surfaces creation by manual selection of boundary points that were 
pictured in more than three images. The generation of the two different surface models and their combination to create a complete 
surface of the monument is described. 
1. INTRODUCTION 
The demand for 3D models of historical monuments is 
continuously growing in the field of archaeological and 
architectural applications. Currently, the two main sources of 
data that can provide detailed and reliable 3D surface models 
are the photogrammetric processing of digital images and laser 
scanning point clouds. 
Photogrammetry is a mature technology for 3D coordinate 
extraction of points, through stereo-restitution or bundle 
adjustment of overlapping images. The number of produced 
points and the level of automation for close range applications 
involving complex objects are still relatively low and require 
time consuming procedures. Although the development of 
specialized software for direct production of 3D models 
following simple photogrammetric techniques have broadened 
the field of applications, yet this does not change the basic 
characteristics of the method. 
The emergence of terrestrial laser scanning and state-of-the art 
software developments for processing the large amounts of the 
produced data may lead to the impression that this technology is 
the main solution for 3D models generation. However, laser 
scanners still remain quite bulky instruments and difficult to use 
especially for data collection entailing higher level positions 
with respect to the ground. Consequently, the existence of 
occlusions and the lack of data in some parts of the objects are 
frequent, resulting to incomplete models. Moreover, although 
there is an autonomy of laser scanning from topographic field 
work, still this is not always the case; i.e. when there is a need 
to incorporate a model into a particular reference system, or 
when the shape and characteristics of the monument do not 
allow a reliable cloud-to-cloud registration, there is need of 
surveying measurements. The use of spherical targets for a 
complete registration of point clouds is often not possible, thus 
increasing the number of acquired scans in the field. 
The combination of products from laser scanning point clouds 
and products of photogrammetric procedures provides a reliable 
result with possibly less field work. This paper describes the 
surface model products obtained from independent 
photogrammetric and laser scanner data and the generation of 
an improved hybrid model using proprietary modelling 
software. The advantages of this approach are discussed through 
a case study of a cultural heritage monument of a Byzantine 
church in Peloponnese, Greece. 
2. SURFACE RECONSTRUCTION USING LASER 
SCANNER DATA 
Surface reconstruction is a well studied problem in computer 
graphics with a wide range of applications. With the advent of 
laser scanner systems, which can provide dense data sets from a 
variety of objects, the issues of surface reconstruction and 
modelling of closed surfaces are receiving great attention as 
they are not completely solved. Moreover, the challenge for 
surface reconstruction algorithms lies in finding methods which 
can cover a wide variety of shapes. 
The main classes of reconstruction algorithms are based on 
spatial subdivision, distance functions, surface warping and 
incremental surface growing (Gopi et al, 2000). The common 
theme in spatial subdivision techniques is that a bounding 
volume around the input data set is subdivided into disjoint 
cells. The goal of these algorithms is to find cells related to the 
shape of the point set. The distance function algorithms define 
the shortest distance from any point to the surface. These are the 
most commonly used algorithms and the approach of Curless & 
Levoy (1996) is well suited to handle very large data sets such 
as those obtained by laser data. Warping-based reconstruction 
methods deform an initial surface to give a good approximation 
of the input point set. This method is particularly suited if a 
rough approximation of the desired shape is already known. 
Finally, the basic idea behind incremental surface construction 
is to build-up the surface using surface-oriented properties of 
the input data points. 
Regardless of the implemented algorithms, Fabio (2003) defines 
four steps for the conversion of the measured data into a