Environmental Engineering Reference
In-Depth Information
TABLE 4.2
Hyperspectral flight lines used for establishment of the image spectral library.
Date
Time
Test
Flight
Flight
Resampled
No. of
heading (
◦
)
(UTC)
site
altitude (m)
pixel size
bands
19.05.1999
10:46
Dresden
179
2540
6.0
128
19.05.1999
12:50
Potsdam
180
2560
6.0
128
01.08.2000
11:50
Dresden
−6
1660
3.0
126
20.07.2003
10:20
Dresden
175
1680
3.0
126
07.07.2004
9:39
Dresden
−4
2050
4.0
126
07.07.2004
10:29
Potsdam
0
1950
3.5
126
20.06.2005
9:35
Berlin
95
1960
3.5
126
17.06.2007
9:30
Munich
0
2520
4.0
126
17.06.2007
9:45
Munich
180
2520
4.0
126
17.06.2007
10:15
Munich
180
2520
4.0
126
25.06.2007
11:05
Munich
0
2520
4.0
126
transformed into a single numerical value for further quantitative
analysis using feature functions, such as absorption position and
depth. Incorporation of such feature values in automated clas-
sification procedures considerably reduces the dimension of the
feature space (Hughes phenomenon) and the variability of input
parameters in comparison to using original reflectance values
as input information for hyperspectral classification methods.
The occurrence of such features is caused by differences in the
intensity of absorption and reflection throughout the reflective
part of the electromagnetic spectrum. The type and wavelength
position of spectral features provide valuable information about
the chemical composition of a material (Clark, 1999). One of
the most important feature types is the absorption band rep-
resented by a local reflectance minimum. The spectrum shown
in Fig. 4.1(d) depicts the absorption band at 2200 nm which
is characteristic for clay minerals, such as kaolinite. Another
commonly used spectral feature type is the increase in reflectance
which can be observed amongst others between the red and near-
infrared wavelength ranges. This so-called ''red edge'' is caused
by the chlorophyll content of green vegetation and described by
(a)
(b)
(c)
(d)
FIGURE 4.1
Exemplary feature functions for the numerical description of spectral features.
Search WWH ::
Custom Search