Image Clustering using a Hybrid GA-FCM Algorithm

Full Text PDF PDF
Author(s) Fagbola, T. Mathew | Babatunde R. Seyi. | Oyeleye, C. Akinwale
Pages 99-107
Volume 3
Issue 2
Date February, 2013
Keywords Image Clustering, Image segmentation, Genetic algorithm, Fuzzy C-means.

Abstract

Image analysis is a process of deriving object description from its image. It is of great theoretical and practical importance in the pattern recognition and image-based security systems domain. Image clustering is the partitioning of an image into the meaningful regions (image classes), based upon the properties of the pixel images, tone and texture. Image segmentation is an important technology for image processing. The goal of image segmentation is to cluster pixels into salient image regions, i.e., regions corresponding to individual surfaces, objects, or natural parts of objects. Segmentation could be used for object recognition, occlusion boundary estimation within motion or stereo systems, image compression, image editing, or image database look-up. A number of clustering algorithms have been applied successfully to image clustering problems among which are K-means, Expectation-Maximization and Fuzzy C-means. This study considers the application of Fuzzy C-means for image clustering. However, the algorithm is sensitive to both noise and intensity heterogeneity since it does not take into account spatial contextual information. To overcome this limitation, GA is introduced to develop a new fuzzy segmentation technique which will optimize the performance of pure Fuzzy C-means. In light of this, a hybrid GA-FCM algorithm is developed to overcome this limitation and help obtain a more accurate clustering output.

< Back to February Issue