In remote-sensing, each image contains different information. Some images give information about the surface of the interested area. Others give information about the underground and mine reservoirs. In some applications, the user may not have the ground-truth of the area, and may not
even know the exact number of classes in the image to be segmented. For this reason, in the presented approach, an elasticity is introduced into the algorithm by making the threshold value
adjustable. The algorithm is capable of producing different segmentation results with different number
of classes. Hence, the user is made free to come up to a decision about the number of classes by
merely changing (adjusting) the threshold value and then, observing the resultant segmented image.
Therefore, it is possible to explore new mine, water or petroleum reservoirs which had not been explored before.