Investigations on clustering based Image segmentation methods for Multi resolution images

Abstract

Satellite images often require segmentation in the presence of newlineUncertainty caused due to factors like environmental conditions poor newlineresolution and poor illumination Specifically in image segmentation newlineproblems the input data involve properties of image pixels sometimes derived newlinefrom very different sources Therefore we need to define different kernel newlinefunctions purposely for the intensity information and the texture information newlineseparately We then combine these kernel functions and apply the composite newlinekernel in Multiple Kernel Fuzzy C Means MKFCM algorithm to obtain newlinebetter image segmentation results newlineThe performance of MKFCM type algorithms depends on the newlineoptimization of segmentation accuracy and its efficiency Based on accuracy newlinethe proposed method is concentrated on obtaining a robust segmentation for newlinenoisy images and a correct detection of small regions based on low threshold newlinevalue The efficiency can be obtained from number of iterations if the newlinenumber of iteration is less the efficiency of the proposed method is good newlineSimulation results obtained for a sample satellite image and low resolution newlinesatellite image requires less number of iterations and segmentation has been newlineachieved for a lesser threshold level newline