4l 2004
————
in area,
and in
bottom
S very
d4-5
'ases to
nizable
lificant
N
ht and
ata are
agnetic
epends
m the
ie first
about
about
height
le and
Xf first
ym 0%
)grams
ms
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
LIDAR intensity data have been processed to produce a
georeferenced 8 bit grey scale raster. Laser data with a density
of 4 points/m’ have been interpolated to give a 20 cm pixel
resolution image (fig. 8), while from data with a density of 1
points/m’ a 1 m pixel resolution image has been obtained (fig.
Figure 8. Detail of laser intensity raster 20cm pixel resolution
Figure 9. Detail of laser intensity raster 1m pixel resolution
Raster images obtained from last echo, in areas covered by
vegetation, have low intensity values of pixel because the
strength of laser power has been lost during several multi-
reflection of laser beam. So rasters represented only with last
echo are not useful to identify objects or entities, but they could
be used to evaluate the vegetative state according to the loss of
laser power passing through the canopy.
As the raster pixel varies according to the point density, the
infrared image obtained from the less dense LIDAR survey
results noisy because the dimension of footprint and the
distance among the laser points make the raster grid size under-
sampled.
However, most of features are clearly recognisable, particularly
detachment surfaces of landslides, roads and fields are clear and
identifiable. This raster image provides a support to understand
ground surface state. The intensity image can be seen as a
secondary product to supplement the primary elevation models
rather than a purely stand-alone product.
The intensity image, thought having similarities to
panchromatic images, does not provide the radiometric
resolution of the satellite orthoimage that is achieved as 16 bit
raster.
949
The raster images from laser data have been used to integrated
the QuickBird orthoimage over that areas that are not clearly
visible because of presence of shadows and clouds (fig. 10).
In the QuickBird image the landslides, particularly the
detachment surfaces, are not identifiable, while, on the contrary,
these objects are clearly visible trough the Lidar intensity.
Moreover the pixel size of lidar intensity image is more detailed
than the pixel size of QuickBird image over Bracigliano area.
So we can integrate QuickBird image with intensity lidar data
and eliminate shadows and clouds over the areas that have to be
studied. At present a simple overlay of the two images has been
performed, but a procedure of fusion is under study.
Figure 10. Particular of an orthoimage and overlay with the
intensity laser data over a landslide detachment surface
A procedure to verify the planimetric differences between the
images is to compare the vectorials of landslides and other
objects obtained from intensity LIDAR data and the other ones
obtained from the different QuickBird orthoimages.
Moreover, the vectorials that describe the shape of landslides
obtained from digitalisation of intensity raster, can be compared
with the contour lines that represent the same value of intensity.
These ones derived from resampling the raster in a new grid
composed by 64 classes of grey scale in accordance with the
belonging to areas with the similar radiometric characteristics
(fig.11). Particularly, around the landslide bodies these contour
lines identify the changement of materials between the bare
rock (sliding surface of landslide) and the undamaged soil and
vegetation.