1SOrS,
Xf the
ance,
vhich
n the
acity.
alies,
)OVe-
st.
tory;
lities
t, the
and
d on
jurth
ring
le is
plex
atia,
oyed
t the
iine
en a
and
azza
irely
ove
and
ome
Orsi
r by
ions
hich
ed a
ania
mal
> IV
1 in
are
ring
30st
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
interesting application in the archaeological field among the
hyperspectral sensors.
MIVIS sensor (Daedalus AA5000) is a scanning system,
founder of new generation of airborne hyperspectral sensors,
working with a high spectral and range gating. It is a modulate
tool consisting of 4 spectrometers taking simultaneously the
radiation reflected by ground surface in the visible (20 bands
between 0.43 and 0.83 jum), in the near-infrared radiation (8
bands between 1.15 and 1.55 pm), in the middle-infrared
radiation (64 bands between 2.0 and 2.5 uum) and in the thermal
infrared radiation (10 bands between 8.2 and 12.7 um) covering
totally 102 bands; besides it is characterized by a 2mrad IFOV
(Instantaneous Field Of View) and 71.1? FOV (Field Of View).
The sensor is housed aboard a CASA 212/200 twin-engine
plane owned by Compagnia Riprese Aeree of Parma.
The aerial swaths over Villa del Casale and Sofiana sites were
taken on 06/20/2002, within a photogrammetric flight plan for
archaeological surveys (fig.1). Two swaths were taken at
different times: the first one at about 9:30 a.m., the second at
about 12:30 p.m., in order to be able to highlight, from the
comparison between the relevant images, the radiometric
variations due to the heating up of the ground. Each swath
consists of two [light strips, North-South direction, 1500 m
height; the images produced are characterized by 3 m geometric
resolution and they are supplied with the only geometric
correction due to the panoramic distortion induced by the
Sensor.
For the image calibration, using a spectroradiometer we could
supply a direct survey, simultaneous to the flight, of the
reflectance values (spectral signatures) of the different ground
typologies in the site (fig. 2).
Figure 2. Direct survey of the reflectance
4. PRE-PROCESSING DATA
In this research the first 28 bands and the last 10 bands were
used. For the first 28 bands the physical datum measured is the
radiant emittance [W/cm’sterad*pum] while the last ten bands
Figure 3. Regional Map overlaid on MIVIS image
supply temperature values obtained from a linear interpolation
between the temperatures of two blackbodies kept respectively
at the temperature of —15°C e +45°C within values above zero.
As it’s known, images must undergo geometric and radiometric
correction. As regards geometric correction, MIVIS images are
affected by distortions connected to the movement of the
aircraft and the sensor characteristics (geometric distortion
within FOV limits). Georeference tests have highlighted the
necessity of using a lot of Ground Control Points and of
splitting up each scene in subscenes with a limited extension to
which to apply top-class polynomial algorithms (Rubber
Sheeting). Anyway, if this approach is satisfactory from a
cartographic viewpoint, it is not satisfactory from the viewpoint
of the following phases of the spectral analysis; in fact, the
resampling of the image pixel during georeference process
modifies the spectral contents. So we have decided to refer the
12:30 image to the 9:30 one, taken as reference image, in order
to resample only one image. Later, after executing all spectral
analyses, the two images were georeferenced to the National
Map System (Gauss-Boaga). In figure 3, the 1:10,000 Regional
Map was overlaid on the georeferenced MIVIS image.
The acquired data are always affected by radiometric errors due
to the sensor characteristics, to the atmosphere and to the
lighting geometry. In the thermal bands these errors are
displayed on the image by stripes.
The correction of these errors and the reflectance calibration are
necessary in order to compare images obtained at different
times, as reflectance is a physical property of the examined
surface.
>
Riflectance
Light object
Dark object
eu
Atmosphric radiance Radiance recorded
by sensor
Figure 4. Calibration line
493
