Introduction - Hyperspectral Image Fusion - page 2

monograph discusses the remote sensing data obtained by the means of variations in
the electromagnetic (EM) energy reflected from the surface of the earth.
Let us take a brief look at the process of EM remote sensing data acquisition. The
sun is the primary source of electromagnetic energy which is propagated through
the atmosphere. Elements of the earth's surface absorb some component of this
incident energy which is partly emitted back, and reflect the remaining fraction.
The amount of energy reflected back into the atmosphere primarily depends on
the composition of the material, and thus varies from material to material. These
variations in the reflected energy are captured by the remote sensing systems that are
attached to either a satellite or an aircraft. This data is then digitized and converted
into a pictorial representation which makes the process of interpretation and analysis
easy. However, before this raw data could be analyzed by the subject experts of
various application fields, it is necessary to convert it into a suitable form. The remote
sensing information is often in the pictorial form, i.e., images. The intensity at each
pixel corresponds to the digitized radiometric measurement of the corresponding
surface area. In remote sensing community, the intensity value of a pixel is often
referred to as the digital number. An ideal image is expected to accurately reproduce
the reflectance of the surface measured within a particular wavelength band. The
intensity value of a pixel is expected to be proportional to the product of the surface
reflectance and the incident energy from the sun (spectral irradiance). However, in
most of the cases, the acquired images suffer from various kinds of distortions. The
atmospheric interference and sensor defects cause incorrect recording of the data
in the sensor system, and thereby introduce radiometric distortion. The factors like
the earth's curvature, and the perspective projection cause errors pertaining to the
shape and size of the objects. These are referred to as the geometric distortions.
Unless such distortions are corrected, these data cannot be effectively analyzed.
Additionally, for different applications, different features of the data are required to
be studied and highlighted. Enhancements of the remote sensing images also play a
major role in the computer assisted analysis. The science of digital image processing
has extensively been applied for the pre-processing of the remote sensing data. The
subsequent analysis and decision making part comprises of the subject experts and
analysts fromvarious application fields. Figure 1.1 illustrates the step-wise schematic
of the remote sensing process.
In this monograph, we discuss one of the important image processing applications
that aims at providing a better visualization of the remote sensing data through
combining multiple images of the same scene. These images provide complementary
information about the scene which when integrated into a single image turn out to
be a quick, efficient, and very useful visualization resource for the human analysts.
When a remote sensing system captures large areas, the altitude of the imaging
system which could be an aircraft or a satellite, increases. Subsequently, this leads to
the reduction in the size of various objects in the scene. The smallest element of an
image, i.e., a single pixel may correspond to an area of a few squared meters in the
actual scene, depending upon the instantaneous field of view (FOV) of the sensor. In
a remote sensing image covering an area of a few hundred kilometers in both length
and width, smaller objects such as buildings and roads occupy a very few pixels