Possiblistic-Fuzzy C-Means Clustering Approach for the Segmentation of Satellite Images in HSL Color Space

Abstract

Abstract Image segmentation is the process of splitting an image into number of sub images or extracting the necessary portions from the image. The segmentation of satellite image is a challenging but important task for the subsequent processes in the image analysis. In this paper, Possiblistic-fuzzy c-means based segmentation of satellite image is proposed. Possiblistic–fuzzy c-means (PFCM) clustering is a blended version of fuzzy c-means (FCM) clustering and possiblistic c-means (PCM) clustering.The PFCM clustering stay away from various limitations of both PCM and FCM. PFCM resolves the noise sensitivity problem of FCM. Moreover PFCM gives answer to the coincident clusters problem in PCM clustering and the row sum constraint problem in FPCM clustering. In the proposed approach, before segmentation, the satellite images are transformed from RGB color space into HSL space. The polar coordinate, user oriented HSL color space approximate the human vision and represents the colors in more perceptually and intuitive manner than the RGB representation. The segmentation of satellite images in RGB and HSL color space is compared and the experimental result shows that competence of the proposed approach.