TABU-ADAPTIVE ARTIFICIAL BEE COLONY METAHEURISTIC FOR IMAGE SEGMENTATION

Abstract

This paper proposes to enhance the Artificial Bee Colony (ABC) metaheuristic with a Tabu adaptive memory to optimize the multilevel thresholding for Image Segmentation. This novel method is named Tabu-Adaptive Artificial Bee Colony (TA-ABC). To find the optimal thresholds, two novel versions of the proposed technique named TA-ABC-BCV and TA-ABC-ET are developed using respectively the thresholding functions namely the Between-Class Variance (BCV) and the Entropy Thresholding (ET). To prove the robustness and performance of the proposed methods TA-ABC-BCV and TA-ABC-ET, several benchmark images taken from the USC-SIPI Image Database are used. The experimental results show that TA-ABC-BCV and TA-ABC-ET outperform other existing optimization algorithms in the literature. Besides, compared to TA-ABC-ET and other methods from the literature all experimental results prove the superiority of TA-ABC-BCV.