ural
the
um,
leal
hic
148
ill
ce
. --
Cu CO "CD
anbul 2004
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part BS. Istanbul 2004
4.1 The Model Formation In table 3 the comparison of the sd. of the point co-ordinates
by 2*1 adjustment with 3Dom adjustment. The low accuracy
of the co-ordinates adjusted with traditional computation, can
be explained with the fact that the theodolite stations are
poorly linked together by reciprocal observations, not using
the suitable equipment. :
The final adjustment with 3Dom has been preceded by the
computation of the approximate co-ordinates with the model
formation. The possible models were the combination of 17
coupled three by three (17/3)=680 and the combination of 17
coupled two by two (17/2)=136 , in total 816 possible
topographic models could be formed; in reality only 27 of
have been created by tri-section (three vertices), the
remaining 29 by bi-section (two vertices). All of them have
been oriented in the same reference system keeping fixed two
vertices of the traverse. In this way the approximate co-
ordinates of the points have been computed. In the following
table the comparison between the co-ordinates by blind
traverse and those got from the final 3Dom adjustment.
Table 1 -Model Co-ordinates /Vs 3Dom co-ordinates (m)
1) sx= .062 sve i0 - sz =.0.034
2) sx2— .008 sy2= .140 sz2= .0.058
3) sxa= .074 sya= .129 sza= .0.0.92
1) sx = sample mean
2) sx2=RMS value
3) sxa=mean absolute values
In figure 6 the differences are shown, amplified 200 times. It
is evident a systematic effect. Figure 7 - The complete network
Table 3 — Comparison RMS value for sd of all points
RMS value (mm) | ox 9, 0;
2+1 adjustment 34 24 14
3Dom 2 2 1
Figure 6. — The differences (amplified 200 times) between
the model co-ordinates and 3Dom co-ordinates-
4.2 The final Adjustments: 2+1 vs 3Dom
Two different types of adjustment have been performed, one
with the traditional procedure where the planimetric
computation takes place before the altimetric one iteratively.
The used software was RETE (Fangi, 1996). The second
adjustment by 3Dom software after removing all the
reciprocal observations between the traverse vertices. 3D
adjustment has the disadvantage to require more computer
memory compared with 2+1 adjustment.
Due to the limited capacity of the computer Ram (512
Mbytes), all the points have been divided in three groups and
the computation performed separately, taking some common
points like the station points and those points with the
maximum frequency. The features of the three adjustment
blocks are shown in table 2
Figure 8. The central part of the control network of the
Guggenheim museum in Bilbao with the ellipsoids of errors
The accuracy improves especially for narrow angle resection
in space. In table 3 the sd. of the adjusted co-ordinates of far | id
Table 2 — The three groups of adjustment station point are shown. i À
- | N points | Equations | Unknown | Redundancy Sd of the co-ordinates of the Far Stations — Comparison |
Group 1 128 812 404 408 traditional 2+1 adjustment /vs. 3Dom adjustment (m) n
Group 2 215 875 648 227 |
Group 3 278 1234 838 396