This research presents study using a satellite image of Landsat TM-5 to obtain a soil
classified map for Northern Sinai using the supervised classification techniques. These
techniques depend on taking the training areas (signatures) on the image from reference
such as geological map for the same area. Three methods of supervised classification
were used, (Minimum Distance, Mahalanobis Distance and Maximum Likelihood) with
different number of classes. Firstly, the maximum number of classes (32) has been used
and then reducing the number of classes in each classification gradually, by neglecting
small areas of classes in the reference soil map in each stage. The number of classes used in these techniques was 32, 28, 24, 20 and 16, respectively using ENVI 4.4
software. The output results of classification techniques have been compared with the
available geological map. The comparison indicates that the method of Maximum
Likelihood gives the best results in all used number of classes. |
