ABSTRACT
In this context, we present a new region-based methodology for the automated segmentation of high spatial resolution images. The proposed method uses the KFCM (Kernel Fuzzy-C-Means) clustering combined to a sequential morphological filtering. KFCM has been adopted because it is robust to noise and outliers and also tolerates unequal sized clusters. We used it to detect the main modes corresponding to the main regions, thus, objects in the image. The proposed filtering may be considered analogous to morphological reconstruction filters. Therefore, the original contribution of the paper is the definition of a morphologic filtering method, which avoids fragmentation or/and over-segmentation. Although the proposed technique may be applied to the study of different types of information, this work focuses on the extraction and segmentation of man-made objects. Indeed, the proposedmethod is particularly well suited for the segmentation of complex image scenes such as aerial or fine-resolution satellite images.Accordingly, the paper is organized as follows. Section II provides a brief description of the segmentation scheme proposed. Examples of segmentation process applied to satellite images are given in Section III. Finally, conclusions and directions for future work are drawn in Section IV.
