Contrast enhancement for satellite image segmentation with fuzzy cluster means using morphological filtering

Abstract

Image segmentation is a stage in image processing, responsible for dividing an image into regions homogeneous, based on the similarity. Suppose the grey level of a pixel by pixel gray neighbors. The quality of image segmentation is generally influenced by the characteristics and the handling of images to be processed. This paper presents the results of satellite image segmentation using a fuzzy cluster region means (FCM). To improve the contrast of the image, conducted morphological filtering techniques. Satellite image analysis performed on five districts in the province of Banten Indonesia. Difference’s segmentation results evident when non-negative parameter value is converted to a 2 and 4. The higher the value of a non-negative parameter is given, then the details of the edges of objects clearer segmentation results. The combined use of a top-hat and boots-hat filtering on objects before satellite imagery analysed by FCM, indicating that it merges with the background object. Background object in the original image is the object of rice fields and is not part of observation in this study. It was identified to have the same gray level similarity with the object of building.