International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
developed programs using points, lines, and combined
points/lines and to test the accuracy of the developed program
using point-based projective equations by comparing its results to
two well-known commercial software programs; Erdas Imagine
' version. 8.5 and Intergraph (Z/I Imaging) I/RASC version 8.4 as
shown in the following subsections.
3.1 Aerial Photography
It was essential to use aerial photographs first, before applying to
satellite imagery, since the projective equations are derived based
on frame geometry. An aerial photograph of a 1:4500 scale for an
urbanized city center was used in this experiment (Figure 1). A
number of 5 control points, and 5 check points were selected to
be used with the point-based projective equations program. Then,
5 control lines, and the same 5 check points were used with the
line-based projective equations program. Finally, 3 control lines,
2 control points and the same 5 check points were used with the
combined point\line program. Table 1 shows the different Root
Mean Square (RMS) results of the 5 check points when using the
projective equations of point, line, and combined point/line. The
results show a superior performance of the line and combined
point/line solutions over the point-based solution for this case of
only 5 control points.
Figure 1. Aerial Photograph of an Urban City Center with
Control (blue) and Check (red) Features Distribution
RMS Point-based | Line-based Combined
Program Program Point/Line
X (cm) 4.28 2.38 2.44
Y (cm) 4.17 2.27 2.18
Table 1. RMS Results of 5 Check Points for Aerial Photography
3.2 LANDSAT7
A LANDSAT7 satellite image was used in this experiment. The
panchromatic band, with its 15-m resolution, was selected
(Figure 2). The image covers an area around Lake Nasr in the
southern part of Egypt.
In the first experiment, the developed program for point-based
projective equations was used with various number of control
points in order to achieve the best possible accuracy limits.
RMS values of 25 check points are listed in Table 2 where it is
clear that 25 control points are needed to achieve the
acceptable accuracy of around 1.1 of a pixel.
&
^. BPO
Figure 2. LANDSAT7 Image of Lake Nasr with its Control
(blue) and Check (blue) Features Distribution
RMS 5 10 15 20 25
control | control control | control | control
points points points points points
X(m) [120.22 | 99.65 65.33 26.15 16.68
X(m) | 121.05 | 99.80 66.03 26.14 16.72
Table 2. RMS Values of 25 Check Points for LANDSAT7
Panchromatic Image Using Point-based Technique with
Various Number of Control Points
Next, the point-based projective equations program was
compared to that of Erdas Imagine and Z/I Imaging I/RASC.
Then, the line-based projective equations program was used
with 5 control lines. Finally, the combined point/line-based
projective equations program was used with 3 control lines and
2 control points. The RMS results of the same 25 check points
are listed in Table 3. It is important to note the equivalence of
the results of the 25 control points with only 5 control lines
and/or 3 control lines and 2 control points. This clearly shows
‘the importance and potential of using straight lines as control
features in the rectification process.
RMS Erdas Point- Line- | Combined
Imagi [RAS based based | Point/Line
ne
X (m) 16.68 16.16 16.68 16.07 16.36
Y (m) | 16.69 16.14 16.72 16.53 16.62
Table 3. RMS Values of 25 Check Points for LANDSAT7
Panchromatic Image Using Point, Line, and Combined
Point/Line Techniques
International .
3.3 SPOT4
Two SPOT4
panchromatic
second is mult
The panchron
developed pro
first to achiev
points was nee
25 checkpoint
RMS
co
pi
Xm) | 11
x an) | 11.
Table 4.
Panchromatic
The comparis
projective equ
performed nex
based and the :
with various
accuracy as wi
values of the
the line-based
for the combin
Figure 3. SPC
Control
RMS | Erd:
mag
X(m) | 13.
Y n) | 13.
Table 5.
Panchron
The second .
multispectral ii
northern part
RMS values o
was needed to
