Thresholding is a simple and efficient technique for image segmentation in digital image processing. In this paper, the fish-swarm algorithm of swarm intelligence is used in image threshoding, and the two-dimensional Otsu thresholding algorithm based on the fish-swarm algorithm is proposed. Numerous experiments show that this algorithm could select the best threshold accurately. In experiments, the threshold, mean and variance from running the two-dimensional Otsu thresholding based on fish-swarm algorithm and single genetic algorithm, as well as the elitist strategy genetic algorithm ten times independently are compared. Moreover, the convergence curves are selected as the evaluation of the algorithm complexity. Experiment results show that this method not only can segment the image more accurately, but also has faster convergent speed.