Image Processing Reference
In-Depth Information
urban growth boundaries from remotely sensed imagery. In the early urban remote
sensing period, mostly film-based cameras (using black and white, color, or color
infrared film) were used to collect aerial photographs, but some studies used video
and multi-spectral digital cameras. Radar, image spectroscopy, and thermal infrared
imaging were also used, but far less frequently. In summary, the scope of early
urban remote sensing was limited to urban planning and infrastructure development
and was hindered by the limited availability of imagery. Since the 1950s, there has
been a noticeable movement from studying urban infrastructure and morphology,
to studying the human dimension of urban environments using remote sensing
techniques.
7.3.2
Recent Urban Remote Sensing (1971 to Present)
This section covers the more recent period in urban remote sensing, 1971 to present,
with an emphasis on the current state-of-the-art. Studies of urban land cover/land
use, urban structure, and change are reviewed. Then, a variety of studies designed
to study urban populations are outlined. Finally, the emergence of a diverse group
of studies concerned with the social science aspects of urban places is traced.
7.3.2.1
Land Cover and Land Use, Urban Structure, and Change Analyses
Studies of urban land cover and land use are the backbone of urban remote sensing.
Although the specific goals of many studies differ, most begin with the identifica-
tion and classification of the land cover or land use within the urban scene. Due to
the many complexities of urban environments researchers have devised new
approaches designed to improve classification accuracies. One of such approaches
has been the development of expert systems (see relevant discussions in Chapter 12);
essentially logic based systems that allows for the integration of remotely sensed
data with ancillary data in an attempt to provide more accurate classification results.
Stefanov et al. (
2001
) developed an expert system to monitor urban land cover
change in Phoenix, Arizona. System inputs included Landsat TM imagery, an image
derived texture measure and vegetation index, water rights and land use data. A nov-
elty of this research is that it was one of the first applications of such an approach
for a semiarid-arid urban environment. The researchers found that improved classi-
fication accuracies support the use of an expert system, especially in urban environ-
ments with heterogeneous land covers. That same expert system was later modified
for use with ASTER data (Stefanov
2002
), and used to study land cover diversity
across a number of different urban centers (Netzband and Stefanov
2003
).
Hybrid approaches to urban land use and land cover mapping have also been
developed. Lo and Choi (
2004
) designed a hybrid method using both unsuper-
vised and supervised approaches as well as hard and soft classifications. They used
Landsat 7 ETM+ data to study metropolitan Atlanta, Georgia. The authors found
