points 
in Fig.1 was ‘imaged 
GEODETIC SERVICE CRC 
f = 240.0mm (changeable), film size =23 
cm. The camera is designed to determine 
precise coordinates of object points by 
simultaneously adjusting with all other 
parameters; interior orientation  parame- 
ters of the camera and exterior orienta- 
tion parameters of photographs (Fraser, 
1982). 
A target field shown 
by a metric camera, 
The field was 4m(height) x Sm(width) x 
2m(depth) in size. 63 target points were 
allocated three dimensionally. Most of 
points were imaged in most photographs. 
Ten photographs were taken, rotating kappa 
by 90 degrees to each other. The order of 
linking photographs adopted in the expeni- 
ment is shown in Fig.3, where the photo- 
graphs 3 and 8 make a datum model, and 
others are linked to this model. The base 
length of the datum model was set to unity 
(1m). 
By applying the procedure mentioned in 
3., relative orientation parameters asso- 
ciated with each independent model coordi- 
nate system were uniquely determined. Al] 
approximations of interior orientation 
parameters but a camera distance were set 
to zeros. The camera distance was set 
initially to 248.5mm, which were read out 
from a micnro-meter-based indicator of the 
camera. No false solutions did not appear. 
In additional experiments the authors con- 
firmed that any other combination of 
photographs than in Fig.3 could make 
models, as long as their convergent angles 
were not near 90 degrees. 
In the case of no control points adjust- 
ment can be done by the method of free- 
network or by the method of minimal con- 
straints. The authors adopted the latter. 
Seven degree of freedom was fixed by 
giving the infinite precision to Zm of 
point a and Xu'YM:7M Of point b,c. in 
Fig.1. As a result of the procedure in 4. 
Table 2 was obtained, which includes the 
approximations and the adjusted values of 
the interior orientation parameters of the 
camera (except for ones related to lens 
distortions) and exterior orientation 
parameters for photo 1 and 10 as well as a 
RMS difference between approximations and 
adjusted values of target point coordi- 
nates. The rotation matreces determined by 
the procedure in 4.3 are both in the form 
[0] [&][K]- 
Table 1 and 2 prove that the algorithm 
produces approximations of parameters 
precise enough for ensuing bundle adjust- 
ments. 
6. CONCLUSION 
This paper discusses the algorithm for 
automatic calculation of approximations of 
parameters in bundle adjustment. Relative 
orientation parameters of each pair of 
photographs are evaluated from the linear 
coplanarity condition. All models are 
linked to form a global model. Then their 
rotation matrices are uniquely decomposed 
to angular elements. If the object space 
204 
the transfor- 
automatically 
coordinate system 
mation parameters 
evaluated. 
is given, 
are also 
The procedure realizes photogrammetry 
without control points or easy orientation 
and camera calibration. It is very useful 
for digital-image-based plotters ( digital 
plotters), which features easy manipula- 
tion for everybody who are not familiar 
with photogrammetry. Actually this has 
been already implemented into a digital 
plotter, TOPCON PI-1000 and now in test 
use. 
The authors would Tike to acknowledge 
Mr.H.O0htani, Mr.M.Chida in TOPCON Corp. 
and Mr.H.Hasegawa and Mr.K.Uesugi in PASCO 
Corp. for their with the 
experiments. 
cooperation 
7. REFERENCES 
+References from Jounals: 
Lohmann,b. ,. G6.Picht, 
K.Jacobsen and L.Skog: 
Development of a Digital Photogrammetric 
Stereo Workstation, ISPRS Journal of 
Photogrammetry and Remote Sensing, 
J.Weidenhammen, 
The Design and 
vol.44,1989,pp.215-224 
Ohtani,H.: Stereo Image Analysis System 
(in Japanese), Image Information, Vol. 21, 
No.25, 1989,pp.49-54 
Frasenr,C.S: 0n the Use of Non-metnric 
Cameras in Analytical Close-Range Photo- 
grammetry, The Canadian Surveyor, vol.36, 
No.3, 1982, pp.259-279 
Arun.k.s., 1.5.Husn. end 5S.D.Blostein-: 
Least Squares of Two 3-D Point Data, IEEE 
Pattern Analysis and Machine 
vol.9,1987,pp.698-700 
Trans., 
Intelligence, 
S.Negahdari- 
Horn,B.K.P.,H.M.Hilden and 
pour: Closed-form Solution of Absolute 
Orientation Using Orthonormal Matrices, 
J.Opt. Soc. Am., A-5, 1988, pp.1128—1135 
+References from Grey Literature: 
Hattori,S5: Calculation of Initia! Values 
of Exterior Orientation Parameters for 
Bundle Adjustment (in Japanese), Japan 
Society of Photogrammetry, 1991, pp.79-82