Image Segmentation Using Teaching-Learning-Based Optimization Algorithm and Fuzzy Entropy

Abstract

Thresholding is one of the most frequently used methods in image segmentation. Fuzzy entropy thresholding approach has been widely applied to image thresholding. Such thresholding approach used two parametric fuzzy membership functions for fuzzy partitioning of the image. In this paper, Teaching-Learning-based Optimization (TLBO) algorithm is used to search an optimal combination of parameters of the membership functions for maximizing the entropy of fuzzy 2-partition. The selected optimal parameters are used to find optimal image threshold value. This new proposed fuzzy thresholding algorithm is called the TLBO-based Fuzzy Entropy Thresholding (TLBO-based FET) algorithm. The proposed algorithm is tested on a number of standard test images. Three different approaches, Genetic Algorithm (GA), Biogeography-based Optimization (BBO), recursive approach, are also implemented for comparison with the results of the proposed approach. From experimental results, it is observed that the performance of the proposed algorithm is more effective than GA-based, BBO-based and recursive approaches.