Abstract: Based on the study of target segmentation in infrared image, a method of infrared target segmentation based on visual cognition was proposed, in order to solve some problems of over-segmentation or lack of segmentation, and insensitivity to the low gray target in gray-scale segmentation algorithm. Firstly, the gray information of infrared image was converted to image variance information; secondly, Chebyshev inequality and some other basic mathematical theories were used to calculate the variance threshold between the target and background, then complete the pre-segmentation of the target information; finally, the morphological filling operator was used to acquire the binary mask image of the target information in order to extract target in the source image successfully. The subjective and objective indices of experiments show that the algorithm of infrared target segmentation based on visual cognition can detect infrared targets in different gray levels effectively，and the misclassification rate of infrared target is very low.