geome- 
cation 
ligital 
’amera. 
. Spec- 
ctural 
le with 
alogous 
|; with 
ted to 
lution. 
sensors 
or are 
Jhmann, 
^ reso- 
hed by 
cases 
entire 
rienta- 
entire 
>xample 
camera) 
réseau 
1ghaus, 
several 
lere as 
known. 
a CCD- 
can be 
r with 
réseau- 
use of 
ras is 
  
2.2 Filters 
For the recording of special object attributes, 
different ranges of the electromagnetic spectrum 
can be selected by using the corresponding fil- 
ters. With this method, it is possible to obtain 
independent images of the different spectral 
regions with high contrast. 
To reach optimal results, it is necessary to use 
filters with steep edges and a limitation of the 
transmission on both sides. These conditions are 
given in metal interference filters with a trans- 
mission range of 20 to 40 nm. In the infrared 
region (»800nm), the half-wave breadth of the 
transmission is even larger. 
A reduction of the half-wave breadth for a better 
selection of object attributes is limited by the 
decreasing transmissivity and is only efficient 
if spectral characteristics of different object 
attributes are well known. 
In the field of art analysis for example, exten- 
sive measurements of spectral reflections of pig- 
ments and binding materials with defined artifi- 
cial illumination were investigated, which 
resulted in the exact knowledge of the spectral 
behavior [Asperen de Boer, 1970]. With this 
knowledge, it is possible to select these spec- 
tral ranges, most useful for defining particular 
object characteristics. In architectural applica- 
tions, especially in the recording of buildings, 
in most cases it is impossible to reach similar 
constant conditions because of the changing illu- 
mination situtation. 
2.3 Evaluation Systems 
For the evaluation of multispectral image data, a 
wide range of software systems is offered, often 
in combination with a link to geo-information 
systems. Most systems are outlined for the evalu- 
ation of remote-sensing data. The software pack- 
ages which are either PC- or workstation-based, 
are able to convert different image data formats 
and allow various image manipulations such as 
image enhancement, image filtering and geometric 
transformation. Additionally, different 
algorithms for unsupervised and supervised 
classification are implemented in most cases. 
These algorithms are well known from different 
publications [e. g. Haberácker, 1989], and will 
not be described here. 
  
Fig.1 Recorded Building 
  
3. EXAMPLE OF A MULTISPECTRAL ANALYSIS OF A 
BUILDING 
3.1 Description of the Object 
The present recording object is the building of 
the "Hochschule für Grafik und Buchkunst" in 
Leipzig (Fig. 1). For this university building, a 
restoration concept is being developed at pres- 
ent. This analysis forms part of an complete 
documentation of the building's condition. A part 
of the Northern front with different building 
materials and typical damages was chosen for the 
examination. 
3.2 Object Recording 
For the survey of the building, three different 
recording systems were selected. Two systems are 
on-line systems: on the one hand the low 
resolution CCD camera Rollei CS1 with an Valvo 
CCD-sensor NXA 1010 with 604 * 576 Pixel (Fig. 
2), on the other hand the high resolution Rollei 
Réseau Scanning Camera (RSC) with approximately 
4200 x 6250 pixel [Riechmann, 1990] (Fig. 3). 
  
Fig.2 Rollei CCD-Camera CS1 
The third system is an off-line system, using a 
Rolleiflex 6006 metric for an analogous film 
recording and a Rollei RS1 réseau-scanner for 
digitising (Fig. 4, Fig. 5) With these systems, 
two different parts of the building were recorded 
with different resolutions, image formats and 
filters as shown in Tab. 1. The digital image 
acquisition with all systems was done with a PC- 
AT386 and software packages, which had been 
developed at the IPB. 
  
Fig.3 Rollei Reéseau-Scanning-Camera