Fig. 2: Panorama acquired with KARLINE 
22 Panorama formation with single frame images 
With this approach custom frame imagery acquired by analog 
or digital cameras is used to compute panoramic views 
(Pomaska, 2002). Using a tripod single images are aquired that 
have to overlap in order to be used for the panorama. A number 
of commercial software products for the production of such 
panoramas are available, e.g. QuicktimeVR (Apple) and 
PhotoVista (MGI). 
The formation of panorama images follows a sequence of 
single steps. Firstly, the images are selected and ordered 
sequentially with respect to their angular position. The field of 
view of each image can be calculated by the given image 
format and the focal length of the applied camera. 
Subsequently, the program generates a panorama segment from 
each individual image (Fig. 3). 
  
    
Fig. 4: Matched panorama segments 
Adjacent panorama segments are matched using image 
information of the overlapping zone (Fig. 4). The matching 
methods are based on similarity measures as known from 
digital image matching. These methods do not provide error- 
free panoramas since they usually do not consider 
geometrically exact photogrammetric camera models. In 
addition the single images are often slightly tilted due to 
mechanical restrictions of the tripod or camera adapter. For 
this reason, the standard routines calculate a randomly 
positioned cutting edge between adjacent images that is based 
on radiometric information inside the overlapping zone. 
These digital blending techniques can generate geometric 
image errors as illustrated in Fig. 5. 
  
Fig. 5: Image error in calculated panorama segment 
—182- 
The 
reco 
ovel 
the 
cost 
pro 
dist 
mar 
the 
clic 
cort 
pan 
moc 
Pon 
In 
pho 
exp 
Th 
WI 
sm 
as 
Zo! 
3.2 
Pri 
ob 
the 
de 
of 
ge