What is urban remote sensing? - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 6

involves field- and image-based spectral investigations to auto-
matically derive quantitative spectral features that serve as the
input information for a multi-step processing system. It allows
detailed mapping of urban surface materials at a sub-pixel level
and provides area-wide information about the fractional cover-
age of surface materials for each pixel. Chapter 5 discusses some
new possibilities and challenges when using very-high-resolution
spaceborne radar data for urban feature extraction. The authors
compare airborne versus spaceborne radar data in terms of image
geometry and other aspects that have been elaborated in connec-
tion to single building extraction, building damage assessment,
and vulnerability mapping. They also discuss the suitability of
adopting the algorithms and methods originally developed for
processing high-resolution airborne radar data to spaceborne
radar data. The last chapter (Ch. 6) included in Part II discusses
the use of lidar remote sensing for three-dimensional building
reconstruction. The chapter comprises a moderate review on
lidar-based building extraction techniques and a detailed dis-
cussion on a comprehensive approach for automated creation
of three-dimensional building models from airborne lidar point
cloud data fused with aerial imagery.

mixture analysis (MESMA) to map vegetation, impervious and
bare soil components. Chapter 9 provides an overview on the
principles of object-based image analysis (OBIA) and demon-
strates how the OBIA can be applied to achieve satisfactory urban
mapping accuracy. Two case studies are conducted with Quick-
bird data to demonstrate two object-based analysis procedures,
namely, decision rule and nearest neighbor classifiers.
The last two chapters included in Part III deal with two impor-
tant aspects for urban mapping: image fusion technique (Ch. 10)
and temporal lag between urban structure and function (Ch. 11).
Chapter 10 reviews some advanced pan-sharpening algorithms
and discusses their performance in terms of objective and visual
quality. Chapter 11 examines the issue of temporal lag between
when decisions are made to change a city to when these changes
actually physically materialize. This seems to be an important
issue for urban mapping. Yet it has been largely neglected in urban
remote sensing literatures. The author explores the temporal lag
largely from a conceptual perspective.
It should be noted that there are some other urban mapping
techniques or methods that have been discussed in other chapters
of this volume. For example, Chapter 2 (Part II) discusses
a hybrid approach combining unsupervised classification and
spatial reclassification that has been successfully used to produce
accuracy-compatible land use/cover maps from a decades-long
time series of satellite imagery acquired by the three Landsat
imaging sensors. Chapter 3 discusses a filtering step built upon
the use of some operators of mathematical morphology as part
of an integrated adaptive spatial approach that can be used
to improve urban mapping from very-high-resolution remote
sensor data. Finally, due to the space limit, we are not able to cover
some other pattern classification techniques that can be used to
improve urban mapping, such as expert systems ( e.g. ., Stefanov,
Ramsey and Christensen, 2001), support vector machines (e.g.,
Yang, 2011), or a fuzzy classifier (e.g., Shalan, Arora and Ghosh,
2003). Readers who are interested in learning more about these
methods should refer to the references provided.

1.4 Algorithms and
techniques for urban
attribute extraction

The urban environment is characterized by the presence of het-
erogeneous surface covers with large interpixel and intrapixel
spectral variations, thus challenging the applicability and robust-
ness of conventional image processing algorithms and techniques.
Largely built upon parametric statistics, conventional pattern
classifiers generally work well for medium-resolution scenes cov-
ering spectrally homogeneous areas, but not in heterogeneous
regionssuchasurbanareasorwhenscenescontainsevere
noises due to the increase of image spatial resolution. Develop-
ing improved image processing algorithms and techniques for
working with different types of remote sensor data has therefore
become a very active research area in urban remote sensing.
For years, various strategies have been developed to improve
urban mapping, and some of the most exciting developments are
discussed in Part III.
The first three chapters in Part III are dedicated to a set of
image processing techniques that can be used to improve urban
mapping performance at the per-pixel, sub-pixel, or object levels.
The first chapter (Ch. 7) discusses some algorithmic parameters
affecting the performance of artificial neural networks in image
classification at the per-pixel level. The chapter comprises a
moderate review on the basic structure of neural networks, two
focused studies with a satellite image covering an urban area to
assess the sensitivity of image classification by neural networks
in relation to various internal parameter settings and the perfor-
mance of several training algorithms in image classification, and
a discussion on a generic framework that can guide the use of
neural networks in remote sensing. Chapter 8 reviews the spectral
mixture analysis (SMA) technique that allows the decomposition
of each pixel into independent endmembers or pure materials to
map urban subpixel composition. It then discusses two case stud-
ies highlighting the flexibility of multiple endmember spectral

1.5 Urban socioeconomic
analyses

Applying remote sensing to urban socioeconomic analyses has
been an expanding research area in urban remote sensing. There
are two major types of such analyses. The first type centers on
linking socioeconomic data to land use/cover change data derived
from remote sensing in order to identify the drivers of landscape
changes (e.g., Lo and Yang, 2002; Seto and Kaufmann, 2003).
The other type of analyses focuses on developing indicators
of urban socioeconomic status by combined use of remote
sensing and census or field-survey data (e.g., Lo and Faber,
1997; Yu and Wu, 2006). While some aspects relating to the
first type of analyses will be addressed in Part VI, here we
focus on some latest developments in the second type of urban
socioeconomic analyses.
Part IV examines some latest developments in the synergistic
use of remote sensing and other types of geospatial information
for developing urban socioeconomic indicators. It begins with a
chapter (Ch. 12) discussing a pluralistic approach to defining and
measuring urban sprawl. This topic is included as part of urban
socioeconomic analyses because defining urban sprawl involves