International Archives of the Photogramme 
carried out. This mode is the most time consuming one and will 
be used in order to fulfil special requirements of a project, e.g. 
thorough mapping of urban street environments. In general it 
will be placed at street crossings where it can also serve as 
valuable tie information for individual facade acquisitions. 
    
Fig.5: CityGrid Scanner at work 
Fig.5 shows the CityGrid Scanner in action driving through a 
narrow street. The scanner is put in horizontal position and 
works in a profile scanning mode (body rotation inactive, 
mirror rotation active). The cameras (their number is theoretic- 
ally not limited) are mounted separately on the vertical device 
bar. which can be moved up to some 4 m in order to look over 
possible obstacles, like parking cars. Though not firmly 
mounted on the scanner body, cameras and laser scanner can be 
kept in a stable relative position to each other at least during 
one acquisition campaign. The arrangement may be calibrated 
in advance or calibration could also be calculated simultan- 
eously later on within the orientation of the entire block 
arrangement. The distance between SGMode stops depends on 
the prevailing situation, but is typically between 20 m to 50 m. 
32 Monument Reconstruction 
Laser scanning has already been approved as appropriate 
method for documentation of monuments. In the following, a 
project should be presented, where a complex sculpture has 
been surveyed with a great number of laser scans and in 
addition with a block arrangement of conventional photographs. 
Scanning and photo shooting has been carried out completely 
independently. The scanning device was a Riegl LMS-Z360 
instrument with a single shot accuracy of some +12 mm, the 
colour camera a Kodak DCS-460c with a 6 MB sensor frame 
and a 28 mm lens. The reconstruction of the object was based 
on a hybrid bundle adjustment, where the intensity images of 
the laser scanner data and photographs have been processed 
simultaneously. In the course of the project the benefit of using 
images together with laser data has been investigated (Haring, 
2003). 
The monument (see Fig.6) has been captured by 22 digital 
photographs and by 20 laser scans altogether (from each of the 
10 positions a coarse scan with 0.2? and a fine scan with 0.05* 
step width was taken). The entire field of interest was targeted 
with 29 signals, one part as retroflecting square foils of 4 cm? 
glued onto the monument, another part as retroreflecting 
cylinders with 5 cm diameter and 7 cm height around the 
monument. They intention was to see from each sensor position 
at least 4 targets, in order to be able to links all captured data to 
each other. 
try, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
  
Fig.6: Marc Anton monument (Above: Kodak DCS image. 
Below: Laser intensity image). Retroreflecting tie point 
targets are numbered. 
For the block adjustment the ORPHEUS/ORIENT software has 
been used. Gross errors in the original scans (especially in the 
coarse laser scans) could be eliminated with the help of rough 
residual analysis, robust estimation, and data snooping. The 
balance of observations vs. unknowns of the joint adjustment 
(laser observation plus photos) is listed in Table 2. 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Number of Tl 
observations in 20 laser intensity images 819 
observations in 22 photos 508 
fictitious observations 6 
intermediate TOTAL 1333 
eliminated after gross error analysis -95 
observations TOTAL 1238 
unknown orientation parameter (6 per sensor position) 192 
unknown coordinates of 30 tie pts (3 each) 90 
datum parameters (3 pts) (3 each) 9 
unknowns TOTAL 291 
Balance = Redundancy 947 
  
  
Tab.2: Balance of observation vs. unknowns in bundle block 
In the course of data snooping a variance component analysis 
has been carried out. The results show clearly what has been 
expected. The variance of the measurements of the coarse scans 
are about 50% worse than those of the fine scan. The accuracy 
of the photo coordinates has been estimated with 0.4 pixels. 
In order to be able to estimate the improvement induced by the 
additional use of photographs, the entire block has also been 
calculated with laser measurements alone. Table 3 lists the 
comparison of the a posteriori accuracies as result of the joint 
adjustment and the sole laser scan adjustment, respectively. 
In this project the usage of photographs was not explicitly 
required because the smooth and complex shape of the 
monument, which hardly bears any texture is a ideal object for 
laser scanner measurements. Photogrammetric compilations 
needed much more effort and probably artificial texturing in 
order to be able to find an appropriate number of points for 
952 
In 
pr 
af 
als 
  
Ta 
4. 
Th 
las 
rea 
det 
obj 
phe 
mo 
sen 
dyr 
pre 
intr 
higl 
atio 
wl 
at li 
ima 
fulf 
of i 
the 
in 3 
lase 
scar 
with 
surf. 
at ol 
mor 
pent 
puls 
Man 
puls 
info 
as ti 
scan 
dayl 
care 
cour 
but t 
alme 
comi 
imag 
Each 
abso 
Mod 
very