3.Description of the Complete Process
Fig. 3 shows the flow of the object based determina-
tion of comparator coordinates and its interdependen-
cy with the orientation. This section describes the
process in detail, Section 3.1 the definition of the
image points, 3.2 their identification.
initial visual measurement
of comparator coordinates and
digitization of the target images
|
image block orientation
with bundle adjustment
|
transformation of photogrammetric
parameters and lens design data
in one unique coordinate system
|
determination of digital target
image models and definition
of photogrammetric image points
image matching of the target
image models with the real,
digitized target images
|
transfer of the photogrammetric
image points into the real,
digitized target images
Significant im-
provement of the image
coordinates?
E: output of bundle adjustment results
Fig. 3: flow of the object based comparator
coordinate determination
3.1.Definition of Image Points
3.1.1.Initial Visual Measurement of Comparator
Coordinates and Digitization of the Target
Images
In this first step initial values for the comparator
coordinates of all target images of the complete image
block are measured visually and the target images
are digitized simultaneously with the measurement.
3.1.2. Image Block Orientation by Bundle Adjustment
The second step is an initial orientation of the image
block, that means, the coordinates of the object
points, the parameters of additional refracting sur-
faces, the exterior and interior orientation parame-
ters are estimated simultaneously by bundle
adjustment (Kotowski 1988, pp.328ff). Since the com-
parator coordinates are not yet sufficient, the results
are not final. But, this initial orientation is required,
because the outcoming parameters are to be used in
the following steps to compute the target image mod-
el. So, the whole process is iterative.
3.1.3. Transformation of Object Points and
Parameters of Additional Refracting Surfaces
into a Geometric Optical Coordinate System
To compute the target image models the coordinates
of the object points, the parameters of the additional
refracting surfaces and the design components of the
camera lens must be available for each image in one
unique coordinate system. For this purpose a geomet-
ric optical coordinate system, typically used in optical
design, is suitable. One of its axes is collinear with
the optical axis; its origin may be fixed in the en-
trance pupil (compare Meid, p. 14).
First the object points and parameters of additional
refracting surfaces are transformed into the photo-
grammetric image coordinate system, which is fixed
in the projection center, using the parameters of the
exterior orientation. The photogrammetric image
coordinate system differs from the geometric optical
one, if
1. the optical axis is not perpendicular to the image
plane,
2. the entrance pupil has another distance to the
image plane,
3. the distortion is not radial-symmetric.
The transformation between the two systems has to
be performed using their relationship to the image
plane, because only the image plane is a physical part
of the camera that is common to both. The trans-
formation equations for the object points are as fol-
lows:
P£? 2RQ(P*-a-b-dH-c)
with
PEP transformed object point in the geometric optical
system
R; rotation matrix, rotating the photogrammetric
system into the geometric optical system;
P* object point, defined in the photogrammetric sys-
tem
a vector, connecting entrance and exit pupil;
b vector, connecting exit pupil and principle point
of collimation;
dh vector, connecting principle point of collimation
and principle point;
c principle distance of the camera.
tior
cal
The
par
are
Tw:
idet
spo:
the
Usi
ima
an
trac
ters
198
ray
trac
tal
ang
wav
one
twee
of tk
dim:
first
poin
mod
poin
If th
a tai
tribu
tion
distr
is es
tion
back
face |
The
the
steps
mode
and
The
are €
in th
1957
distr
into :