International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
Where K is the ratio between the Root Mean Square
Error (RMSE) and the pixel dimension ( 0.2 <K < 1.5)
- For the two used couples, control points were
determined by aerotraingulation based on classic
stereopreparation. The accuray, in planimetry and
altimetry, of points determined by stereopreparation is
estimated about + 20 cm and the one by aerotriangulation
is about £60 cm. Thus, the corresponding formula may be
given:
n,x0.20+n,x0.602 |"”
eame ]-——— 0)
points obtained by
those obtained by
n, is the number of control
stereopreparation and n;
aerotriangulation.
- The planimetric error on terrain coordinates depends on
the pixel dimension and the scale of negative. This one is
given by :
t
Gp = (oto)? = V2 x — x Ec (4)
2
Where t is the pixel dimension and Ec the scale number .
- The altimetric average error depends mainly on the
pixel dimension, the scale of the negative and on the ratio
base — height (B/Z). It is given by the simplified formula
which follows :
t Z
x mm (5)
2 B
Where Z represents the diffeérence (H-h) between the
altitude of perspective center (H) and the point of interest
(h).
Gr V2 x Ec
- The accuracy of absolute orientation depends on
altimetric and planimetric acuracy of scanner coordinates
and the accuracy of control points. Hence, the planimetric
accuracy is given by :
as 2 21/2
Opt (Op tk Oa )
And the altimetric accuracy is given by :
2 aU
oz 7 (07 * 04)
3. EXPERIMENTAL STUDY
3.1 Case of AL Hoceima couple
3.1.1 Interior orientation
Table 2 shows that the calculated pointing error, either
for left or right photo is less than the theoric error which
is half pixel. Hence, measures can be juged as
acceptables for the continuation of the process.
366
Resolution (dpi) 300 400 600 800
Pixel dimension 85 64 42 32
(um)
Pointing | Left 36 19.92 14.76 9.44
error | photo
(um) | Right 21 18.67 8 5
photo
Theoric pointing
error 42 32 21 16
Table 2 : Interior orientation results
3.1.2 Relative orientation
Table 3 shows that, generally, the differences are less
than half pixel. We can also notice an enhancement of
relative orientation with increasing resolution.
Consequently, the process can be continued and goes to
the absolute orientation.
300 400 600 800
RESOLUTION
dpi)
Taille du pixel 85 64 42
m
o, (um 49 17 10 6
K 0.57 0.26 0.24 0.19
Table 3. Relative orientation results
3.13 Absolute orientation
Table 4 provides absolute orientation results that is
planimetric RMSE (mxy) and altimetric (mz) calculated
for the different resolutions. Moreover, this table
provides also the corresponding theoric accuracies in
planimetry (cp) and in altimetry (7). By analizing these
results, we can notice that, for any used resolution, the
planimetric RMSE is out of tolerance. However, for the
altimetry, we can notice that the errors are aceptables for
300 and 400 dpi resolutions but not for those of 600 and
800 dpi. This result may be due to the terrain topography
and vegetation cover which may have an effect on
features identification. We can notice also accuracy
enhancement with increasing resolution.
Resolution 300 400 600 800
(dpi)
mxy (m) 4.259 3.129 1.865 1.830
Gp (M) 2.448 1.863 1.2936 1.021
mz (m) 2.905 2.398 1.931 1.817
o; (m) 3.618 2.727 1.855 1.245
Table 4. Absolute orienation results
The obtained results urged us to use the resolutions of
600 and 800 dpi Sidi Slimane couple. These results are
discussed next.
3.2 Case of Sidi Slimane couple
3.2.1 Interior orienation
Table 5 shows that the calculated pointing error, either
for left or right photo is less than the theoric error which
Internati
is h
acce[
Res
Pixe
Point
errc
(pn
The
Table
one f
qu'en
pixel.
orient
proce
orient
Resc
Pixe
323
Table
planir
for t
provi
planir
table
in pla
these
the pl:
Resol
n
A
Before
shoulc
sont b
the ter
turn m
3.3 Re
For th
Were (
for the
points
Furthe
