 This paper proposes a new global multi-level thresholding method for image segmentation. It utilizes that Salis entropy, which is a generalization of the Shannon entropy, to determine optimal thresholds. Additionally, it employs the artificial B colony, ABC, algorithm to quickly identify the best thresholds. Experiments show that the proposed method outperforms other methods in terms of accuracy and speed. This article was authored by Yudong Zhang and Lenin Wu.