In view of the disadvantages of salp swarm optimization algorithm, such as low optimization accuracy, easy to fall into local optimum, and K-means algorithm for image segmentation easily disturbed by the initial cluster center, an improved salp swarm optimization K-means algorithm was proposed for image segmentation. Firstly, circle mapping was used to initialize the salp population. Secondly, Levy flight was introduced into the leader and follower position updating formula to improve the diversity of salp population and overcome the algorithm falling into local optimum. Finally, eight benchmark functions were used to test the performance of the improved salp population swarm algorithm. Then, the improved salp swarm algorithm is optimized with K-means for image segmentation. The improved algorithm improves the searching accuracy, stability, convergence speed and the ability to jump out of local optimum. At the same time, the K-means algorithm was optimized by improving salp swarm algorithm to improve image segmentation quality effectively. The improved algorithm improves the disadvantages of the original salp swarm algorithm, such as low optimization accuracy and easy to fall into the local optimum, and can effectively optimize the K-means algorithm for accurate image segmentation, which has a certain reference significance in the field of image segmentation.