Enhanced image segmentation using hybridized optimization techniques

Abstract

Digital Image Processing has found its application in a number of domains ranging from Medical Diagnosis Military Surveillance Operations Remote Sensing and Criminology etc The raw image is fed to the Digital Image Processing system It is processed to obtain the result required by the end user Of the various steps involved in Digital Image Processing Image Segmentation plays a vital role It is the process in which the image is simplified to locate objects and boundaries so that it could be used for further processing to obtain a result A large number of models and algorithms to solve the problem of Image Segmentation have been proposed However there are a lot of issues in the identification of an accurate segment This thesis addresses the problem of accuracy to a certain extent by employing four different algorithms in the process of segmentation The first one being the application of Exponential Particle Swarm Optimization is one of the variants of the Particle Swarm Algorithm This variant uses an exponential factor in the inertia weight to solve the problem of stagnation and makes the segmentation process faster This algorithm on deployment over a number of images has been found to be more efficient compared to that of a number of previously used algorithms for segmentation newline newline