ıl 2004
P
Se
Css
ted
ble
in
ese
the
the
for
ind
hy
on
1CY
ia =
of
are
her
ich
is half pixel. Hence, measures can be juged as
acceptables for the continuation of the process.
Resolution (dpi) 600 800
Pixel dimension 42 32
(um)
Pointing | Left 6.08 4.12
error photo
(pm) | Right 4.47 5.39
photo
Theoric pointing 21 16
error
Table 5: Interior orientation error
3.2.2 Relative orientation
Table 6 shows that in general the differences are less than
one fifth of a pixel. D'aprés le tableau 6, on remarque
qu'en général les écarts sont inférieurs au cinquiéme du
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.
Resolution (dpi) 600 800
Pixel dimension 42 32
(um)
o, (uum) 6 5
K 0.14 0.16
Table 6. Relative orientation results
3.2.3 Absolute orientation
Table 7 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 (op) and in altimetry (c7). Moreover, this
table provides also the corresponding theoric accuracies
in planimetry (cp) and in altimetry (07). By analizing
these results, we can notice that, for any used resolution,
the planimetric RMSE is less than the theoric error.
Resolution (dpi) 600 800
mxy (m) 0.896 0.797
oP (m) 1.342 1.082
mz (m) 0.487 0.276
6Z (m) 1.897 1.477
Table 7. Absolute orientation results
Before going to the step of restitution evaluation, we
should emphasize that results for the case of Sidi Slimane
sont better than those of Al Hoceima. This may be due to
the terrain topography and the vegetation cover which in
turn may have an effect of features identification.
3.3 Restitution evaluation
For the purpose of restitution evaluation 11 GPS points
were colected for the case of Al Hoceima and 14 others
for the case of Sidi Slimane. The coordinates of these
points were adjusted based on known planimetric points.
Furthermore, taking into consideration both resolutions :
367
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
600 and 800 dpi, the coordinates of these points were
compared with their corresponding points provided by
restitution. Table 8 presents the obtained average
differences during this verification.
Resolution (dpi) 600 800
Al D,, (m) 3.90 1.996
Hoceima | D, (m) 6.93 2.033
Sidi D,, (m) 3.00 1.890
Slimane | D, (m) 2.07 2.011
Table 8. Results of restitution verification
By analizing table 8, we can notice that for the same
couple, the results is enhanced from resolution 600 to 800
dpi. We can also notice that the obtained result for Sidi
Slimane is better than the one for Al Hoceima. This may
be due to the terrain topography which is rough in the
case of al Hoceima and flat in the case of Sidi Slimane
Moreover, to conclude for the scales of maps to be
produced, we have considered the tolerated differences
which are used by the National Cartographic Agency in
Morocco. These differences are + 0.2mm at map scale for
the case of planimetry and + 1/3 of contour interval for
the case of altimetry. Therefore for a map at scale of
1/50000, we obtain 10 m in planimetry and 3.33 in
altimetry.
By taking into account these differences, those of the
collected points by GPS and their corresponding points
collected from restitution, we can conclude from the
results of table 9 that topographic mapping at scale of
1/50 000 is possible using aerial photographs at scale of
1/40 000 and an A3 office scanner.
Resolution 600 800 600 800
(dpi)
% des points
à écart < 10m
% des points
à écart < 3.33 m
Al 82% 100% 27% 91%
Hoceima
Sidi 100% 100% 35% 71%
Slimane
Table 9. Results of restitution verification
4 Conclusions
This research has dealt with the study of accurac of
digital restitution from images acquired by means of an
A3 office scanner. The purpose was to see at which level
can restitution results be used for topographic mapping at
medium scale.accuracy Considering that the scale of
photographs is 1/40 000 and the resolutions are 300 and
400 dpi, we are faced with a problem of features
identification and hence the results are not acceptables.
On the other hand, for the resolutions of 600 and 800 dpi,
the results are better and apply to topographic maps
standards at scale of 1/50 000 and lower. Thus, it is
possible to conclude that an A3 office scanner with a
resolution of 800 dpi is sufficient to convert photographs
at scale of 1/40 000 to images for the purpose of a digital
