Image Processing Reference
In-Depth Information
The optical spectral range is separated into the visible
(VIS: 400-700 nm) and near infrared (NIR: 700-2,500
nm). The NIR is commonly divided into two broad regions
based on detectors used to measure radiation, including
the photographic NIR (700-1,000 nm), a region that can
be sampled by photographic film or silicon-based detectors
and the short-wave-infrared (SWIR: 1,000-2,500 nm),
which requires other detector materials. In general, imag-
ing spectrometry supports our spectral understanding of
different land surfaces and thereby improves our ability to
discriminate materials using remote sensing.
Remote sensing of urban environments is particularly challenging. The land
surface objects (e.g., buildings/roofs, roads) have a small spatial extent. Given this
large amount of spatial heterogeneity most analyses in urban areas have relied
upon aerial photography as a data source. Recent advances in spaceborne systems,
such as IKONOS and QUICKBIRD provide cost effective alternatives to aerial
photography. For example, IKONOS provides 4 m multispectral data, thereby
meeting the minimum spatial resolution of 5 m considered necessary for accurate
spatial representation of urban materials such as buildings and roads (Jensen and
Cowen 1999 ). But urban environments also possess a high spectral heterogeneity
(Ben-Dor et al. 2001 ; Roberts and Herold 2004 ). They are characterized by a large
diversity of materials such as human-made features, vegetation, soils, and others.
Since spectral understanding is essential for remote sensing applications, the fol-
lowing sections will focus on this issue and give an introduction to the spectral
dimension in urban environments. Based on principles of spectrometry, we discuss
the spectral characteristics and the spectral discrimination between urban land cover
types. We show examples of remote sensing mapping applications to highlight the
effects of different sensor configurations on the urban mapping accuracies.
hyperspectral
systems sense the
earth surface in a
large number of
narrow spectral
bands for precise
identification of
the chemical and
physical material
properties
4.2
Spectral Characteristics of Urban Surfaces
The generic study of reflectance characteristics or spectral signatures is usually
based on spectral libraries. These libraries contain pure spectral samples of sur-
faces, including a wide range of materials over a continu-
ous wavelength range with higher spectral detail, and
additional information and documentation about surface
characteristics and the quality of the spectra (i.e., meta-
data). They can be derived from laboratory and ground
spectral measurements, and from hyperspectral remote
sensing observations. Figure 4.2 highlights this with
example spectra of urban surfaces. In field spectrometry
the sample measurement is taken along with a calibration
signal of a 100% reflectant material. The ratio of both
spectral libraries
include spectral
samples of
surface materials
derived from lab-
oratory, ground
measurements or
hyperspectral
imagery
 
Search WWH ::




Custom Search