EE AE ET ERA RES IE APE IR
FER AURI
eA
Y— coordinate of the projection centre
indicate that for high-precision problems
the pre-calibration of the focusabie
cameras does not satisfy the accuracy
requirements. It is necessary to use the
method of simultaneous calibration in
which the elements of the internal and
external orientation are determined
together using photos of the object and
of control points suitably distributed in
the object space simultaneously with the
solution of the photogrammetry problem.
4. The elimination of the lens distortion
stabilizes the values of €f and Ys
obtained from the different photos and
improves the accuracy. Thus before
calibration a correction of the image
coordinates with a pre-set lens
distortion is recommended instead of
increasing the number of the control
points necessary for the analytical
determination of the lens distortion.
SIMULTANEOUS CALIBRATION
The established differences in the
analytically determined values of f and
Ys, obtained from different photos
motivate the necessity for simultaneous
calibration for each photo, taken with a
focusable camera. If for the purpose of
improving the accuracy of the solution of
the photogrammetric problem a double or
triple fixing of the object state from
each camera station is done, simultaneous
calibration and determination of the
external orientational elements for each
photo is necessary. To avoid averaging of
the photogrammetric results from all
photos a method for the simultaneous
adjustment of the data obtained from the
multiple photography of the object from
two or more camera stations with a
simultaneous calibration and
determination of the external orientation
elements. The calibration and the
determination of the external orientation
elements can be done by common or by
individual elements for the photos taken
from the same station.
The analytical solution is based on the
collinearity condition and aliows a
simultaneous adjustment of the data from
multiple photography of the object from
two or more camera stations. Each photo
participates independently in the
adjustment, and the internal and external
orientation elements are determined
simultaneously with the solution of the
photogrammetric problem. The method is
analogous to the block phototriangulation
by bundle adjustment.
Computer program
On the basis of the method mentioned
above a program is created which allows
the accomplishment of following:
(i) pre-calibration of cameras by a test
object -with or without determination
or elimination the lens distortion
(ii)calibration of photographs in the
process of solution of the photogram-
metric problem:
- with camera calibration elements
common to all photos, taken from
all stations;
- with orientation elements common
to ali photos, taken from one
station;
- with camera calibration elements
different for each photo;
(iiijcalibration of all or of only some
camera parameters;
(iv) only a correction of the measured
values of the camera external
orientation elements obtained by a
pre-calibration.
Experiments
The method and the program described
above were experimented on a test object
photographed from two camera station with
three photoplates each. The following
cases were experimented:
ti) for different number of unknown
elements of internal and external
orientation:
6 elements -the external orientation
elements only,
7 elements - the same 6 elements
plus the focal lenght,
9 elements - the same 7 elements
plus the prinsiple point
coordinates of the photo,
10 elements - the same 9 elements
plus the first radial
distortion coefficent,
712 elements - the same 10 elements
plus the second and third
radial distortion coeffi-
cients,
(iij in the case of simultaneous
calibration and determination of the
orientation elements, using one, two
Or three photographs from each
station (simultaneous adjustment);
(iii)in the case of different number and
distribution of the control points.
The standard errors and RMS errors
obtained for some of the experimental
cases are give in Table 1.
The photogrammetrically obtained spatial
coordinates of the points are compared
with the geodetically determined ones.
The root mean square errors allong each
coordinates axis have been computed from
their differences and used as criteria of
accuracy of the solution. The following
accuracy can be considered as realized:
nm 1,5 mm , m, - 5 mm , m, = | mme