Image Processing Reference
In-Depth Information
images. Image variation between dates signals the occurrence of land cover change
(Martin
1989
).
Comparison of Classified Imagery
:
A study was conducted in North Eastern Cairo,
in Egypt, using multi-source data, Landsat-TM, Spot-HRV and KVR-1000, with
spatial resolution of 30, 20 and 5 m, respectively. Overall, the merged images
TM-KVR and HRV-KRV provided higher information content. It was possible to
map out land use changes over 48 years (1945, 1986, 1991, and 1993), which offers
a valuable indicator for urban growth (CEDARE
1998
). Two similar studies were
conducted respectively in the cities of Casablanca (Benchekroun
1993
) and in
Rabat-Salé (Benchekroun and Layachi
1993
).
Combination of Classified Imagery
:
A methodology was developed by Mongkolsawat
and Thirangoon (
1990
) to develop land cover change detection methodology for the
Yasothon province of Thailand. Two Landsat-TM imagery acquired during the wet
and dry seasons (September and April respectively) were co-registered and classi-
fied based on land cover and terrain types. The September image was classified into
11 classes, while the April image was classified into 14 classes. The two classified
images were then mathematically combined resulting in a unique linear combina-
tion of 154 possible gray levels image. Through a process of regrouping these gray
levels into classes, the resultant image provided a meaningful land cover dynamics
with respect to the terrain types.
Comparison of the Classified Radar Imagery
:
Smara et al (
2003b
) describes a radar
based change analysis methodology in which two Radar SAR images of ERS satel-
lite were acquired for the periods of 1992 and 1996. The techniques proposed were
applied and the results were compared against the principal component analysis
method. The quantitative study revealed some changes in buildings, vegetation and
other related seasonal variations (Smara et al.
2003b
).
Comparison of Conventional Data
:
Johnson (
2003
) investigated the impact of land
degradation along a coastal strip in Togo, both under the urbanization encroach-
ment and water erosion. The study produced two land use maps, based on 23 aerial
photographs at a scale of 1:10,000 taken in 1985 and the topographic map of 1960
at a scale of 1:50,000. These two maps were analyzed by GIS. Changes of a period
of 25 years were found to be related mainly to the retreat of the shoreline (ampli-
tude ranging from 45 to 230 m depending on the locations) and the great entropic
pressure on the other land use categories, especially the coconut palm.
13.3.2
Land Use Change Analysis and Modeling
Land Use Analysis and Modeling
:
The integration of time series econometrics and
artificial neural networks improves the monitoring of change using a series of
images. This kind of time series analysis typically requires an understanding of the
drivers of change. This understanding could be built using a multivariate regression
