International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B6. Istanbul 2004 
  
  
Combined Many points Bad repartition of 
matching feature points, 
clusters of points on 
textured area 
Wrong matching on 
problematic images 
Regular Many points Wrong matching on 
combined Good repartition of | problematic images 
matching feature points 
  
  
  
  
  
Table 1. Results from the five methods. 
These tests lead to several conclusions described in the 
following table. More generally, they give encouraging results, 
even on problematic images (repetitive structure for example). 
Moreover, the experience shows that the methods do not have 
the same efficiency according to the nature of the scenes. And 
increasing the number of generated points corrupts the quality 
of the measures. While the implementation of a suitable 
algorithm is expected, the method to be applied must be chosen 
case by case. 
  
Figure 2. Position of the new generated points. 
We can see on fig. 2 the distribution of the points calculated 
according to the used method: 
a) Iterative feature-based matching 
b) Combined matching 
c) Feature-based matching 
d) Regular combined matching 
The experimentation was done on an Etruscan amphorae 
coming from the excavation of the Grand Ribaud F. 
3.6 Conclusion about DSM 
Among the numerous possible applications of this process, the 
completion of laser measurement as presented is very useful. To 
survey places where laser instrumentation is too difficult to 
settle, classical photogrammetry is a good compromise to get 
automatically a dense network of measures. 
4. DATA STORAGE AND PUBLICATION 
4.1 General description of the data and goals 
ARPENTEUR allows the user to produce data relative to a 
photogrammetric survey. Information coming from the 
knowledge of the measured objects and the site is added with 
these data. The communication between measured data and 
information on the survey is of primary importance in such a 
system, and provides a synthesis of all information available in 
a survey. It makes it also possible to check the results in a 
visual way, and allows an automatic check of the data 
coherence. The format of data storage must thus answer these 
concepts, but it must also be integrated perfectly into the 
ARPENTEUR package. For that it must allow the publication 
of data on the Web, but also to be simple of installation and use. 
4.2 XML, a standard for data storage 
The data used or produced by ARPENTEUR belong to three 
groups: photogrammetric statements, information on the survey 
and measures. The backup of these data thus requires the ability 
to store sorted information in classes. XML format makes it 
possible to create files that gather the information sorted out of 
tree. This is conceptually interesting. Moreover, from a 
practical point of view, the XML is a standard resulting from 
World Web Consortium (W3C) very widespread in the medium 
of research and in companies. You can see the W3C 
recommendation in (XML, 2000) XML is particularly 
appropriate for our application. XML has also other interesting 
characteristics. First of all, a file XML is a simple textual file, 
what allows an edition and a very simple distribution. Then, a 
XML file is readable without having to make specific operation 
(contrary to the file of data bases for example). More, the 
textual nature of the file makes it sensitive to compression. As 
XML format is introduced, we can describe its specific use in 
ARPENTEUR. 
4.3 XML: the core of data processing 
The choice made for the development of ARPENTEUR is to 
limit to the maximum the redundancy of information. The data 
are all stored in a single way. The use of these data then implies 
the treatment of XML files, being used as common core to all 
the chain of use and treatment of the application. Data are 
stored in three XML files. The first one contains the 
photogrammetric model of the survey, the second one contains 
expert knowledge necessary to the survey and the third one the 
photogrammetric statements and measurements, like all 
information relating to the survey. For example, for the survey 
of the excavation of a wreck, the first file contains information 
relating to the calibration of the diving apparatus and the 
orientation of the photographs. The second file contains 
knowledge to measure on the objects (amphorae...). The third 
contains the results of the photogrammetric statement, the 
measurements of the objects as well as the remarks and 
information given by the archaeologists. From these three files, 
all information concerning a survey is available. 
4.4 Simple data publication: XSL 
A transformation of the XML documents using XSL style 
sheets is carried out. See example on the W3C web site (XSL, 
1999). This transformation makes it possible to visualize the 
majority of the data in a static way, without any possibility for 
the user of interacting with those. This representation allows a 
126