Abstract: In order to solve the problem of low efficiency and poor stability of traditional optimal entropy threshold segmentation.A novel fruit fly optimization algorithm was proposed and the image segmentation threshold was optimized.Gaussian sampling was used to update individual fruit flies. In the early stage, due to the scattered distribution of fruit flies, the chance of jumping out of the local extreme value could be increased.At the later stage of optimization, the population distribution of fruit flies is relatively concentrated, and more accurate searching can be carried out.In addition, make full use of the iterative results of the previous generation of fruit flies, generate learning factors, and guide the optimization of offspring fruit flies.Experimental results show that the improved algorithm has made great progress in solving efficiency and precision, and achieved better results than other algorithms in the application of image segmentation.