Comparative Analysis of Differential Evolution Algorithm Using Shannon, Fuzzy, and Cosine Similarity Entropy Functions for Satellite Image Segmentation

Abstract

This paper illustrates a comparative analysis of thresholding based novel technique for satellite image segmentation. Differential evolution algorithm is implemented for multilevel thresholding of colored satellite images using Shannon, Fuzzy, and Cosine similarity entropy functions. Segmentation is a vital step and prerequisite for understanding satellite images so that the information can be utilized in several application areas such as crop monitoring, forest monitoring, transformation changes on land usage, disaster and crisis support management systems. The evaluation of the implemented algorithm is carried out with the help of statistical measures such as MSE, PSNR, and SSIM.