In order to avoid the early convergence of the original whale optimization algorithm and easily fall into the local optimum, firstly, the cat mapping generation chaotic sequence combined with the reverse solution method was used to replace the randomly generated initial population in the process of initializing the original whale optimization algorithm. Secondly, the crazy operator and the golden sine algorithm were used in the position updating mechanism. Finally, the improved whale optimization algorithm was used to find the maximum entropy of 2-D image to determine the optimal threshold of image segmentation. The simulation results of 10 classical benchmark functions show that the original whale optimization algorithm can increase the initial population diversity and search ergodicity, and improve the global search ability and get rid of the local optimum. The results show that the optimal value of the function is 0, 0.00030, -3.32.The improved algorithm can achieve accurate segmentation of target image and less time consuming. This study provides a reference for the application of group intelligence algorithms to image segmentation.