International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
and of control materials, according to working aims.
3.1.1 Medium-resolution satellite image: Viewing from the
government demand on macro strategy in land resource,
medium-resolution satellite image, such as TM, CBERS,
SPOT2/4, IRS and KOMPSAT images can be optional. Refers
to the tablel, better choice would be TM and SPOT2/4 with
lower cost, better image quality, robust programming function
and good service.
Resolut Revisit ; ;
t ; Swath Price
Data ion Period (Kım) (Scene)
(m) (Day)
TM/ETM 15/30 16 185 380/5008
SPOT2/4 10 2-3 60*2 1000$
CBERS 20 3-5 113
IRS-1C/ID 5.8 S 70
KOMPSAT 5 2-3 17
Table 1. Comparison of medium-resolution satellite images
:50,000 geographic map with 10m planimetric precision and
:50,000 DEM with precision of 4-19m (plat-mountain zone) in
elevation are the good auxiliary materials for rectification of
TM/ and SPOT2/4.
3.1.2 High-resolution satellite image: With high ratio of
capability-price as table2, SPOTS 2.5m image is preferred to be
selected for updating 1:10,000 landuse map among the various
data sources of SPOTS, IKONOS, Quickbird and
aerial-photo,etc.
Data Resolution Reve Swath Price
su (m) P. (km) (km? )
SPOTS 2.5/10 2-3 60 2.41€
IKONOS 1/4 2-4 11 30$
Quickbird 0.6/2.8 1-6 16 308
Table 2. Comparison of high-resolution satellite images
GPS control point with precision in cm level is the first choice
as GCP for rectification of SPOTS, except that, 1:10,000
geographic map can be also optional.
1:10,000 DEM is good auxiliary material for image
rectification, especial for image with mountain zone, but it is
difficult to obtain widely, so 1:50,000 DEM can bc as the
substitute for rectification of SPOTS.
3.2 Processing of RS data
Rectification model, GCP number and distributed rule, fusion
method can be introduced separately in this chapter.
3.2.1 Rectification model: There are several models for image
rectification: physical model, polynomial model and rational
function model.
Physical model: Constructs the strict model with the
parameters of satellite and sensor, it requires DEM and fewer
of GCP, can greatly decrease the workload and increase the
precision of rectified image. PCI, Erdas,and other softwares can
provide physical models for SPOT and TM/ETM.
Polynomial model: Simulates terrain form only with GCP, no
needs of the parameters of satellite and sensor and DEM. This
model can be optional for plain zone.
Rational function model (RFM): Helps accessing to high
precision by simulating camera model without parameters of
orbit and camera. It is just for the satellite that can’t supply the
parameters but with complicate model.
The following table 3 lists the first two models in different
softwares for satellite images of TM/ETM and SPOT:
Software TM/ SPOT | SPOT | Quikbird
ETM 2/4 5 IKONOS
PCI9.0 1+2 142 142 3
ERDAS6.0 1+2 1+2 l 3
GEOIMAGE4.0 l 1+2 1+2
ENVI4.0 1+2 1+2 j+2 3
. Table 3. Software and models (1-7 Physical model,
2-Polynomial model, 3-RFM)
3.2.2 GCP number and distribution
GCP number and distribution seriously rely on rectification
models.
Physical model: In theory, 16 GCP is enough for TM and 6 for
SPOT; but in fact, number of GCPS is higher with different
terrain conditions (plat.vs.mountain) and different softwares,
for ex, 12-20 GCP is better for SPOT.
GCP is necessary to give a well-proportioned distribution and
control four edges in image, just as in figure2 (red marks).
Figure2. distribution of GCP
Polynomial model: GCP number varies greatly with terrain: it
is enough to have about 10 GCP in plat zone, but in mountain
more, the number should be up to 20-50 GCP.
It is necessary to give an equal distribution for GCP in rectified
zone; besides, in order to increase mosaic precision, common
GCPS are needed in the joint zones of different rectified
ranges.
3.2.3 Rectification and Match: In this article, rectification is
special for the geometric processing of pan image (P), and
match is for Multi-spectral image (M).
GCP, topographic map can be control materials for the
rectification of pan image, and rectified pan image can be good
reference for Multi-spectral image.
In general, pan image is firstly proceeded, then Multi-spectral
image. But for the data with same acquiring condition (same
satellite, orbit and VOF, etc.), for ex, SPOT5 programming data,
it is possible to match Multi-spectral image to the raw pan
image at first, then merge them, finally rectified fusion image
with control materials. Rectified images will keep their original
resolutions.
Internation
Hubner?
In order to
indices are
RMS of G!
geometric C
produced W
value varie
GCP.
Geometric:
it is checke
GPS or othe
3.2.4 Imag
enriched w
important t
from many 1
Image com
for fusion o
P&M image
get a good «
good if reso
mark in tabl
~ p
fs |
TM/E
SPOT
SPOT
In this projec
popular com
Image fusia
resolution |
resample to t
General fusi
HIS and Bro
the different
Figure 3-5 li
obvious that
construction.
