Image segmentation algorithm is sensitive to intensity of illumination. The different intensity images are not usually segregated availably based on one kind of algorithm, and therefore in this paper a method of image segmentation based on Spiking Cortical Model (SCM) is proposed. Firstly, choosing different stages membership function to approximate characteristics of dynamic thresholding attenuation. Secondly, choosing the different criterion function to suit for different intensity image based on the difference of bright image and dark image. The Experiments show that proposed algorithm could obtain the segmentation result availably and more details information of image. And it has the stronger robustness, which is better than the improved pulse-coupled neural networks、the improved Intersecting Cortical Model (ICM) and two-dimension OTSU.