Image Processing Reference
In-Depth Information
Hyperspectral imaging has been found to be highly useful for a wide span of
application areas. Every object on the earth's surface possesses a specific pattern of
energy reflection across the electromagnetic spectrum. Hyperspectral data have two
unique characteristics—a very fine spectral resolution, and a large number of bands.
This advancement of imaging sensors and the growing power of computers have
enabled researchers to explore various applications of hyperspectral data. During the
initial phase of research, the hyperspectral data were mainly used for surveillance
purposes by the military. These images reveal spectral characteristics of the objects
which can potentially be used to uniquely identify them. Due to their classification
ability, hyperspectral images have been used in defense and military to detect various
camouflages. It could be used to detect hidden objects where employing a conven-
tional RGB camera fails due to its poor spectral resolution. It is also possible to
make use of hyperspectral imagery to detect a camouflage by the enemy from the
vegetation or the soil. Automatic spectral target recognition is an important military
application of hyperspectral data. Traditional methods of target detection are based
on the spatial recognition algorithms. With the advent of hyperspectral imagery, it
has become possible to reveal the spectral characteristics of the target resulting in a
better target identification [152]. The HYMEX project of the Canadian defense has
been intended towards detection of military vehicles, camouflages, and man-made
materials [6].
Some of the bands in hyperspectral data are affected by various atmospheric
constituents. These include water vapour, carbon di-oxide, aerosols, etc. The analysis
of these bands brings up the information about the corresponding factors affecting the
band. For example, the analysis of the bands near 0
.
µ
94
m is useful for the detection
of the presence of water vapour [153].
Hyperspectral data provide an accurate information about the scene and its com-
position. It has proved to be a rich source for various geo-exploration activities. The
remote sensing images can be used to explore minerals, soil, snow, etc. Minerals are
usually found beneath the earth's surface in a pure or a mixed form. The sensors
AVIRIS and Hyperion have demonstrated the ability to remotely map the basic sur-
face mineralogy [93], while the data obtained from the HyMAP hyperspectral sensor
have also been analyzed for the purpose of mineral exploration along with accurate
map generation, and soil characterization [74].
The hyperspectral data are capable of revealing various characteristics of the
soil such as its type, moisture content, and erosion. With the help of readymade
soil spectral libraries, it is possible to determine the material decomposition of the
soil [6]. The fertility level of the soil and its variation over a spatial neighborhood
within a field can be investigated with the aid of hyperspectral data [75].
The list of application areas is continuously growing. Typical applications include
environmentmonitoring [157, 170], characterization of land and ocean areas [42, 129],
characterization of ice and snow areas [53, 121, 154], crop and vegetation assess-
ment [38, 117], etc. Hyperspectral data have also proved to be quite useful for med-
ical applications [56, 161, 179], and industrial applications [69, 171]. Readers are
requested to go through these references for detailed information. The focus of this
