International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004
RMS Errors on the Control Points [m
Scene No.
X
01 +1.8
02 z 1.9
03 +1.9
04 +24
All Scenes + 1.6
tA | oc
Y
+2
+ à.
= 2.
=
zd
ol
RMS Errors on the Check Points [m
Y
20
3.2
ais
+22
Z
im [1
to
bi E Mel
2.1
Table 3. RMSE of the orientation results of block scenes HRS SPOT 5
4. THE METHODOLOGY OF DTM GENERATION ON
THE BASIS OF PANCHROMATIC HRV SPOT 4 DATA
4.1 Specification of methodology
The methodology of measurement of DTM on the
panchromatic images HRV SPOT 4 was elaborated. The
modules of photogrammetric software’s like MATCH - T
Inpho and Image Station Automatic Elevation Collection
(ISAE) Z/I Imaging were used for measurements of grid
of elevation points with rows and columns spaced
dependent from terrain type.
The basic factors influencing the results of correlation
measurement of DTM are described in Ph.D. thesis of
Mr. Ewiak. The most important factors are:
- . Radiometric accuracy of SPOT 4 images;
- Accuracy of delimitation of Rational Function
Coefficients for stereoscopic SPOT 4 Pan scene;
- Method of correlation;
- Parameters of the software used for auto-
correlation measurement of height-points.
Close correlation was found between accuracy of DTM and the
radiometric accuracy of SPOT 4 images. In order to assure
maximum effectiveness of correlation measurement, the
optimum radiometric processing has been done on the raw
panchromatic SPOT 4 scenes.
It has been also found that the accuracy of delimitation of
Rational Function Coefficients, responsible for the correctness
of exterior orientation parameters of SPOT 4 scenes, influences
fundamentally the accuracy of measurement of DTM.
It show also, that the incompetent usage the added parameters
of software realizing the correlation measurement of the DTM
can demote the accuracy of height measurements.
The problem of limitations of MATCH - T as well as the ISAE
was solved. This problem concerned the admissible size of
measuring grid of DTM for stereoscopic SPOT 4 images.
4.2 The results of measurement of DTM
The measurements of DTM were performed on the
separated fragments (“Region I” and "Region II") of
stereoscopic scene SPOT 4. The rational function
coefficients were calculated in independent process.
For the "Region I" the grid of height-points was generated with
spacing 25m and for "Region Il " with spacing of 100m. The
accuracy of measured DTM were examined on the basis of their
comparison with reference DTM, which were obtained by auto-
correlation methods from aerial photographs in scale 1:26 000,
with manual edition.
The accuracy of reference DTM was estimated as RMS Z = +
0.5m. The results of analyses of accuracy for each DTM, taking
into account the influence of the factors described above is
shown in table 4.
The use of elaborated methodology, permitted to measure
DTM with accuracy RMS Z = + 4,7 m for mountainous and
hilly terrain and RMS Z = + 2.6 m for flat terrain.
Additionally, Mr. Ewiak has elaborated software for filtration of
wrong DTM points and their removal, as well as the
interpolation of new points on the basis of the analysis of
neighboring terrain. The software is written in programming
language Borland Delphi 6.0 for the operating systems
Windows NT, 2000, XP. Using this software the accuracy of
DTM was increased to RMS Z = £3.9 m for mountainous
and hilly terrain and to RMS Z = + 1.8 m for flat terrain.
The factors influencing accuracy of correlation measurement of DTM
RMS on height points of DTM
for different test regions [m]
Region I Region II
Raw images of SPOT 4 + 8.8 *4s
Radiometric accuracy of images + 8.4 t4.]
Accuracy of delimitation of Rational Function Coefficient r786 + 3.4
Selection of software parameters t 6.8 t 2.6
The grid size of DTM + 4.7 +26
Table 4. The influence of basic factors on results of measurement of DTM by correlation method of the HRV SPOT 4
[nter
—
Epip
imag
matc
have
16-b
only
inter
Gen
gros
in ta
For
rang
imag
The
mea
perr
acci
pan
date
Tec
ins
be ı
Afr
can
and
To
basi
suit
add
etc.
