The main idea of this paper is to adapt the Artificial Bee Colony metaheuristic to solve the problem of multilevel thresholding for image segmentation. More precisely, this method is exploited to optimize two maximizing functions namely the between-class variance (the Otsu's function) and the entropy thresholding (the Kapur's function). This leads, respectively, to two versions of the ABC metaheuristic: the ABC-Otsu and the ABC-Kapur. The robustness and proficiency of these two thresholding algorithms are demonstrated by applying them on a set of well-known benchmark images. Furthermore, the experimental results show the efficiency of these two thresholding methods.