Remote Sensing and GIS for Land Use …
151
Geometric correction addresses the errors in the relative positions of pixels. The
process of geometric correction and rectification involve the use of several wellknown ground control points (GCPs). These GCPs include road intersections, railways, buildings, etc. to be matched in the original rectified map and the image to
be rectified. The images are then rectified to the UTM projection system (Universal
Transverse Mercator).
The image was rectified to the Universal Transverse Mercator (UTM) Zone 35
north with datum of WGS 84.
C. Image sub-setting
The images were investigated and it was found that the data set cover not only the
study area, but also a great part of the western desert. Therefore, ENVI 5.1 software
was used to subset the image by the vector that include the boarder of the study area.
This process decreases the amount of digital data in order to speed up processing
which is important when dealing with multiband data.
Image processing
Satellite image classification
Image classification could be defined as the automatic process of classifying
all the pixels of a digital image into particular land cover classes (Lillesand et al.
2003). Supervised image classification technique is a user-controlled process where
pixels are given into specified land use classes based on pre-determined locations
that collected previously from field, aerial photographs, and maps (Jensen 1996).
Support Vector Machine (SVM) classifier was used in the current study for classifying different remotely sensed images. Comparative analysis that was done by
Devadas et al. 2012 clearly revealed that the object-based SVM method resulted
in overall classification accuracy (95%), while it was less using traditional pixel-based
classification (89%)
The post-classification change detection analyses describe and quantify the
changes that occured in the same scene at different times. Change of land use classes
area between three classified images was calculated using the post classification
process. According to Hegazy and Kaloop (2015), the post classification analysis is
very useful to identify the types of changes that happened in different classes of land
use such as decrease in agricultural land or the increase in urban built-up area and
so on.
4 Results
4.1 Main Land Use/Land Cover Classes
An up to date land use/land cover map was produced based on the supervised classification (SVM) for multispectral Sentinel 2 image dated 2018. The studied area was
151
Geometric correction addresses the errors in the relative positions of pixels. The
process of geometric correction and rectification involve the use of several wellknown ground control points (GCPs). These GCPs include road intersections, railways, buildings, etc. to be matched in the original rectified map and the image to
be rectified. The images are then rectified to the UTM projection system (Universal
Transverse Mercator).
The image was rectified to the Universal Transverse Mercator (UTM) Zone 35
north with datum of WGS 84.
C. Image sub-setting
The images were investigated and it was found that the data set cover not only the
study area, but also a great part of the western desert. Therefore, ENVI 5.1 software
was used to subset the image by the vector that include the boarder of the study area.
This process decreases the amount of digital data in order to speed up processing
which is important when dealing with multiband data.
Image processing
Satellite image classification
Image classification could be defined as the automatic process of classifying
all the pixels of a digital image into particular land cover classes (Lillesand et al.
2003). Supervised image classification technique is a user-controlled process where
pixels are given into specified land use classes based on pre-determined locations
that collected previously from field, aerial photographs, and maps (Jensen 1996).
Support Vector Machine (SVM) classifier was used in the current study for classifying different remotely sensed images. Comparative analysis that was done by
Devadas et al. 2012 clearly revealed that the object-based SVM method resulted
in overall classification accuracy (95%), while it was less using traditional pixel-based
classification (89%)
The post-classification change detection analyses describe and quantify the
changes that occured in the same scene at different times. Change of land use classes
area between three classified images was calculated using the post classification
process. According to Hegazy and Kaloop (2015), the post classification analysis is
very useful to identify the types of changes that happened in different classes of land
use such as decrease in agricultural land or the increase in urban built-up area and
so on.
4 Results
4.1 Main Land Use/Land Cover Classes
An up to date land use/land cover map was produced based on the supervised classification (SVM) for multispectral Sentinel 2 image dated 2018. The studied area was