International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
1.1 Study Area And Data Used
Büyükçekmece (41° 03' latitude, 28° 45' longitude) is one of the
32 administrative districts, which comprise the city of Istanbul.
It is surrounded by the Marmara Sea on the south, the east by
Avcilar and Kügükcekmece, the north by Catalca, and the west
by Silivri districts (Figure 1). It has 9 administrative sub
districts: Büyükcekmece Centrum, Kavakl, Yakuplu, Kirag,
Gürpinar, Esenyurt, Mimarsinan, Kumburgaz and Tepecik and
also has 5 villages (Türkoba, Hosdere, Cakmakli, Karaagaç and
Ahmediye). The population in the district is generally higher in
coastal sub districts due to high rates of migration, and in
summer this rate is maximized because of the summer
residences.
Figure 1. The map and satellite images of the study area. (a)
Ikonos image dated 13/ 02/2002 (b) QuickBird image dated
03/05/2003.
In this study, very high-resolution images of Ikonos and
QuickBird-2 dated 2002 and 2003 in panchromatic band were
used as remotely sensed data. Besides, the digitized 1/5000-
scaled orthophotos dated 1996; Leica System 300 GPS receiver
and Garmin eTrex summit pocket GPS data were used to
evaluate the geometric accuracy of the images. The
specifications of the remotely sensed data used are given in
Table 1.
Satellite Date | Spectral Res.(um) | Spatial Res.(m)
0.45-0.52 (Visible)
0.52-0.60 (Visible)
Ikonos 2002 | 0.63-0.69 (Visible)
0.76-0.90 (NIR)
Pan 0.45-0.90
| m Pan
4 m MS
0.45-0.52 (Visible)
0.52-0.60 (Visible)
QuickBird2 | 2003 | 0.63- 0.69 (Visible)
0.76-0.90 (NIR)
Pan 0.52-0.90
0.61-0.72 m Pan
2.44-2.88 m MS
Table 1.Specifications of the data used.
1.2 Methodology
In this study, the analysis of the geometric accuracies of these
two very high resolution images were conducted by the
geometric rectification using with rapid-static GPS survey,
pocket survey and digitized 1/5000 scaled orthophotos.
As a first step, the panchromatic band of Ikonos and QuickBird
were enhanced using different digital enhancement techniques
for better visual analysis, such as contrast modification and
filtering.
Rapid static GPS survey was the base information source for
this study. While two stable GPS receiver were collecting data
on the known stations, which one of them is IGS-ISTA
permanent station in the Istanbul Technical University and the
other is a triangulation point determined in the study area; one
rover GPS receiver was used to measure the coordinates of the
GCPs selected. Measures of X, Y and Z coordinates were
recorded at 5 sec intervals for about 5 minutes with at least four
satellites availability for each point. Most of the observed points
are wharves, buildings and parcels corners. Then, these
collected GPS data were transferred to computer and their UTM
coordinates were calculated. UTM coordinates of GCPs were
calculated about cm accuracy by using Ski-Pro GPS data
processing software. The second field survey was conducted
with pocket GPS and UTM coordinates of the selected GCP's
were gathered and recorded automatically. As à third data
collection method, the same GCPs were selected from the
digitized 1/5000-scaled orthophotos. All the coordinates
obtained from different techniques calculated in UTM
projection system in ED50 (European Datum 1950).
2. APPLICATION AND RESULTS
When considering 2D transformation models, many models can
be checked such as polynomials, similarity, affine and
projective models. In this research, 1% and 2'" degree
polynomials model was used due to its simplicity and
availability within most of remote sensing software packages.
Two orders of polynomials were utilized to check the metric
integrity of images. The two transformation models adopted for
testing were as follows (Shaker et al., 2002). 1st and 2" order
polynomials;
X, = b,+b, X+b5Ÿ (
Y, = a,+a,X+a5 Ÿ (
X,» b, b; X-tb, Y-b,XY-b X^ b, Y^ (
Yo= ayta,X+az Y+ayXY+asX*+ag Y” (
1018
Int
W|
Or
of
me
ass
GC
Sq
for
del
me
ha
we
Car
an
Fig
