Under the ocean background, a segmentation method was proposed based on the fusion of multiple features for infrared ship segmentation. This method is used to extract horizontal edge information and vertical edge information from infrared image. First of all, the mean of the data collected on different scales was got, and the result image was seen as the first feature. Furthermore, for the problem of ship targets with different sizes, the improved multistage filters was employed for infrared image, so as to prohibit background and highlight target. The multistage filtered image was identified as the second feature. Finally, the local maximum gray value of infrared image was identified the third feature. Then these three features would be normalized and integrated. During the process of infusion, firstly, each featue image was given weight; then, an appropriate infusion coefficient for fused image was selected. The fused image was segmented by adaptive threshold, followed by a morphological plastic in order to remove isolated areas, supplement holes and improve the segmentation results. The simulation results show that, compared with traditional segmentation strategies, the proposed segmentation method based on multiple features fusion is more likely to meet the demands of segmentation.