Satellite image segmentation using RVM and Fuzzy clustering

Abstract

Image segmentation is common but still very challenging problem in the area of image processing but it has its application in many industries and medical field for example target tracking, object recognition and medical image processing. The task of image segmentation is to divide image into number of meaningful pieces on the basis of features of image such as color, texture. In this thesis some recently developed fuzzy clustering algorithms as well as supervised learning classifier Relevance Vector Machine has been used to get improved solution. First of all various fuzzy clustering algorithms such as FCM, DeFCM are used to produce different clustering solutions and then we improve each solution by again classifying remaining pixels of satellite image using Relevance Vector Machine (RVM classifier. Result of different supervised learning classifier such as Support Vector Machine (SVM), Relevance Vector Machine (RVM), K-nearest neighbors (KNN) has been compared on basis of error rate and time. One of the major drawback of any clustering algorithm is their input argument that is number of clusters in unlabelled data. In this thesis an attempt has been made to evaluate optimal number of clusters present in satellite image using DAVIES-BOULDIN Index.