sis features 
It is highly 
the results 
ntation 
lual models 
nt. It trans- 
ters into a 
ring instru- 
imeters into 
p the opera- 
arallax free 
S are consi- 
le necessary 
ore it is as- 
] on an ana- 
idard device 
ments. The 
it is consi- 
iges will be 
ter. Measur- 
ng. For this 
meters com- 
used in the 
odel. 
connected to 
ring process 
ns. All pro- 
henever it is 
is given in 
leasurements 
vs the opera- 
ts and mea- 
red to the 
is holds true 
ich serve as 
n are image 
refer to..the 
ally given by 
any case a 
tablished in- 
ents are car- 
d coordinate 
ng device to 
1e before the 
This type of 
r+ orientation. 
s transforma- 
the accuracy 
only fiducial 
format noth- 
1 between the 
> camera is 
formation for 
each individual mesh can be applied. This leads to a 
much better modelling of film deformations. The 
reseau correction is image related and not camera 
specific. This means that the effect of local film 
deformation is modelled by the inner orientation 
and not by the bundle adjustment. Furthermore the 
film deformation varies with time, temperature and 
humidity. But all these effects can be properly 
modelled by use of a reseau and be eliminated by 
the inner orientation. 
During a photogrammetric project the inner orienta- 
tion program is used very often. After the film has 
been placed on the stages, the inner orientation pro- 
gram will guide the operator through the whole 
measuring process. The program will drive to pre- 
defined locations and wait until the operator has 
measured the point. Depending on the amount of 
measurements already made, the distribution of 
points in the image and the preselected type of 
transformation the program will constantly compute 
as many transformation parameters as possible with 
the current set of data. The user interface will keep 
the operator informed about the status of the mea- 
suring process. He can see the points already mea- 
sured the ones still to be measured distinguished by 
shape and color. This feature is mainly of interest 
for reseau measurements. Furthermore the residuals 
after the transformation are displayed on the screen 
for inspection by the skilled operator. The user in- 
terface of the measuring programs looks very much 
the same as of the examples given in chap. 5. 
At any time the operator can interact with the pro- 
gram through the graphical user interface. He can 
mark points to be excluded from the adjustment of 
inner orientation parameters or need to be remea- 
sured. Also he can change the type of transforma- 
tion at any time. 
After measurement and adjustment of inner orienta- 
tion has been finished, the operator will accept the 
result, which will then be transmitted to the ana- 
lytical plotter for further use. 
3.2.1. Measurement of Reseau before Triangulation 
As the reseau correction is image related, it needs 
to be repeated each time the image was removed 
from the measuring instrument. To save time during 
the measuring process only those points should be 
measured which are really required for the current 
purpose. Triangulation is a process which works 
pointwise. That means there is no information need- 
ed in between two points. Therefore the measure- 
ment of the reseau can be limited only to those 
meshes which surround the points used in the trian- 
gulation process. 
3.2.2. Measurement of Reseau after Triangulation 
After the bundle adjustment has been carried out 
and the images have been put back into the ma- 
chine, the inner orientation needs to be redone. But 
in case of a reseau camera now only those crosses 
need to be measured which cover the area of inter- 
est. It might happen that the crosses to be measured 
do not match the ones used for triangulation. But 
this is not important. As the operator wants to drive 
through a parallax free model after bundle adjust- 
ment, these reseau corrections need to be trans- 
mitted into the analytical plotter for on-line 
correction of film deformations. If the operator 
docsn't know in advance exactly which area he will 
use later on he can measure the whole reseau, which 
will also take a few minutes on an analytical instru- 
ment only. 
If the operator wants to do selective measurements 
of the reseau the graphical user interface makes it 
very easy to select the area of interest. After the 
selection the program will guide the operator 
through the measuring process. 
4. BUNDLE ADJUSTMENT 
Traditional bundle programs for aerial applications 
are optimized for the aerial case and also based on 
assumptions concerning the block geometry. The 
process of taken photographs for map making is 
highly standardized. The bundle programs can rely 
very much on the assumption that photographs were 
taken according to the standard situation. Typically 
a block is formed by strips with regular overlap 
within strips and between neighboring strips. The 
geometry is quite simple, because usually the image 
plane is parallel to the object surface. The types of 
observations in aerial case are normally limited to 
image coordinates and coordinates of control 
points. 
However the situation in close range applications is 
quite different compared to the aerial case. In close 
range applications, specially in industrial applica- 
tions, the geometry of a block varies from project 
to project. The types of observations used in close 
range applications include also geodetic observa- 
tions like distances besides the normal photogram- 
metric observations. Furthermore as the cameras 
used in close range applications are quite different 
compared to aerial cameras, the cameras very often 
need to be modelled in a certain way. As the block 
geometry isn't standardized as in the aerial case, 
prediction of accuracy in the object becomes more 
difficult. 
This type of applications requires a specialized 
bundle program to fulfill the various requirements. 
4.1. CAP 
CAP (Combined Adjustment Program) is a bundle 
adjustment program which was specially designed 
to fulfill the needs of close range applications. 
4.1.1. Types of Observations 
CAP is capable of handling following types of ob- 
servations: 
* image coordinates 
* model coordinates 
* Spatial distances 
* horizontal distances 
* directly observed coordinates (weighted control 
points) 
* coordinate differences (height differences)