Segmentation of images captured by the satellite system is a challenging task due to non-uniformities in illumination and the features present in an image. Segmentation problem can be modelled as a problem of selecting appropriate multi-level threshold values for maximizing the image entropy or between-class variance. These goals are incorporated in Tsallis, Otsu’s and Kapur’s objective functions. An effective optimization algorithm is needed to optimally search the distinct threshold values in the direction of maximizing the objective functions for that image. In this methodology, Ant Lion Optimiser (ALO) is employed to search the optimal threshold values. A comparative study is carried out with the existing literature for exhibiting the applicability, adoptability and superiority of the proposed ALO methodology for satellite image Segmentation. This approach is tested on different satellite images and the results are analysed with four performance metrics namely PSNR, MSE, SSIM and FSIMC indices. Solution of ALO algorithm with Kapur’s entropy function as objective produces promising result for satellite image segmentation.
Volume 11 | 02-Special Issue
Pages: 570-586