|
Recently, due to the advance of image scanning technology,
imaging spectrometry scanners which have tens or even
hundreds spectral bands have been invented. Images have such
a large number of bands are usually called "hyperspectral
images". Comparing to the traditional multispectral images,
hyperspectral images include richer and finer spectral
information than the images we can obtain before.
Theoretically, using hyperspectral images should increase our
abilities in classifying land use/cover types. However, when
traditional classification technologies are applied to process
hyperspectral images, people are usually disappointed by the
consequences of low efficiency ,needing a large amount of
training data, and barely improvement of
classification accuracy.
Currently, how to use hyperspectral data efficiently is still
unclear. In order to solve this problem, this research focuses
on the data analysis of hyperspectral images. First, form the
viewpoint of data representations, this thesis illustrates the
characteristics of three different spaces (Image, Spectral
and Feature) in which hyperspectral data canbe inspected, and
introduces some fundamental statistical theories and graphical
presentations of the statistics for hyperspectral images. Then,
spectral differences are analyzed in difference spaces. In
addition to spectral differences, differences between
two hyperspectral data classes are also analyzed.
Because of the high correlation among hyperspectral bands, we
use Principal Component Transformation and Fourier
Spectrum Transformation to remove the spectral redundancy
and calculate the class separabilities for the transformed data,
such as the Euclidean Distance and the Distance. After
transformation, we can obtain stable class separabilities in
the low dimensional feature space, so that we expect
that traditional classification techniques can be properly
applied to transformed data.
|