The remote sensing society is currently offering a wide variety of digital images that cover most of
the Earth's surface. The up-to-date image data is a promising tool for producing accurate maps. To
maximize the benefit of such data, automatic and efficient classification methods are investigated.
For the past years, traditional pixel-based classification has been used. Currently, a recent
classification concept, object-based classification, has been prospected. The recent concept's basic
principle is to make use of important information (shape, texture and contextual information) which
is only in meaningful image objects and their mutual relationships. The main aim of the present
work is to find the most suitable technique from the available ones for feature extraction which
can be applied for Egyptian environment.
For this study, high resolution satellite image from IKONOS satellite was used to carry out the
image classifications. The ground reference data were collected from field observations and
personal knowledge. At the present work, the methodology focuses on comparing between two
classifications techniques through application on four test areas with different specifications with
respect to its planning. The first technique is the traditional pixel-based image analysis and the
second one is the object-oriented image analysis. Software ERDAS V.9.2 was used for pixelbased
image analysis and classification. The object-oriented image classification was performed
through eCognition Developer software V.8.0. Accuracy of each one of both techniques was
evaluated through overall accuracy and kappa coefficient from the error matrix and then
compared to each other. Results of this work showed that object-based image analysis has more
advantages than the Pixel-based one. Also, it is found that as the more planned area as the higher
results accuracy.