International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
precision data is increased. In this way one is able to make use of
the same information obtained before processing but at a greater
degree of resolution. Before applying this method however, it is
necessary to correct radiometrically and to geo-reference the
images in order to make the relationship between the single
pixels univocal (figure 2). If however the data has the same
spatial resolution, a simple co-recording can be carried out. This
procedure allows for an orientation of one of the two images in
relation to the other, independently of the reference system in
which the data is furnished. HSV allows merging of RGB type
images with panchromatic images, therefore only a limited
number of bands (three with this method) may be employed. The
different sensor channels are therefore associated with the three
basic colours in order to generate an false colours image
representing a physical reality, such as might be, for example,
temperature distribution in the upper atmosphere or the rate of
ground level humidity. This operation requires a good knowledge
of the data one wishes to use and great care in the choice of
bands to associate with the three RGB channels. The principle on
which this algorithm is based is to separate and extrapolate,
starting from the false colour image, the information relating to
the intensity (I) to the saturation (S) and to the hue (H). From a
strictly mathematical point of view, the system of coordinates (I,
S, H) can be thought of as an Cartesian orthogonal system; given
that the human eye treats the three components as if they were
bound by an orthogonal type relationship (Bretschneider and
Kao, ), and can be written (Pohl, 1998) as follows:
Lua 3 ond
41 ive BB Ir
dp QM
Vil-|-; 7 ET G
y. Ys ys Jo | 4
La
X: DN
(1)
Hz m
V,
5= y Gy
&
The variables Vi and V2 are only and exclusively used to
calculate H and S and have no direct connection to the image.
The first of the three coordinates, that represents the intensity
(1), is directly related to spatial position in that its value is bound
to the type of surface analysed, while the coordinates H and S
supply additional radiometric information by describing its
composition and degree of absorption. In figure 2 the sequence
of operations to be carried out is shown synthetically.
Once the information is deconstructed, it is possible to substitute
one of the three components with the information contained in a
fourth ‘channel, which has not undergone any kind of
transformation. Generally it is the I component to be substituted
with the intensity value obtained from the higher resolution
image. Once this operation is completed, one proceeds to invert
transformation (1) so as to obtain again an RGB type image.
The formula that allows for such a passage is the following:
A
=
|
N
—
v
Sl
Before starting transformation (2), the information regarding
saturation and colour (S, H) is opportunely re-sampled by using
linear type techniques and stretching techniques in order to adapt
the multi-spectral bands to the band furnished by the highest
resolution panchromatic sensor. HSV, therefore, is an evolution
of the IHS algorithm, which has now become a standard in the
procedure of Image Analysis.
[impu 1| p En
; fs Radiometric correction | :
| Geometric correction | ;
| Collateral Data
Figure 2. Flow diagram of operations
3. SENSORS UTILISED
The MODIS sensor (Moderate-resolution Imaging
Spectroradiometer), was installed by NASA. together with other
typologies of instruments, for the analysis of the atmosphere, the
oceans and land above sea level on Terra and Aqua launched
respectively in 1999 and 2001. These are two of the most used
satellites for global monitoring of the environment. Figure 3
shows the diagram of the MODIS sensor.
MODIS is a spectrometer capable of observing the Earth, with a
return period equal to 1 — 2 days and using 36 channels (see table
4). The spectral bands utilised are included between 0.45 and
14.4 um (21 bands included in the interval of 0.45-3.0 um, 15
included in the interval of 3.0-14.4 pm) and furnish a spatial
resolution at the nadir, depending on the band, of 250 m, 500 m
and 1 Km.
Thanks to the high number of channels and to short employment
frequency, this sensor is used for various applications, among
which the following:
- Determining different scales of cloud cover
- Assessing Aerosol concentration
- Assessing plant cover and soil productivity
- Estimating snow masses
- Measuring soil temperature
- Monitoring fires
722
Inter
Sp
Por
Doub
Scan
The /
prede
and
Atmc
mete
envir
cycle:
veget
1979
with 1
NOA
spect!
equal
frequ
atan
AVH
