Morphological profile based segmentation for satellite images

This paper presents the use of morphological profiles to understand the context of the scene of concern. There are many studies in the literature exploiting morphological information focusing on granulometry and anti-granulometry. The well-known morphological feature vectors are the morphological profile and its derivative, MP and DMP respectively. These studies mainly consider the most significant scale of the profiles as of geometrical information for the pixel under consideration. On the other hand, some of the studies holistically treat the whole profile as a feature vector for classification or hierarchically analyze the profile to select the best representative scale. In this paper, we investigate the relevance of the MP and DMP to segment VHR images. We presented the results of four different methods applied to QuickBird images of rural and urban areas taken from the city of Ankara, Turkey. Experimental results demonstrate the capabilities of these methods along with their limitations.