For the pores edge detecting of the Activated Carbon Fibers (ACF) material images, traditional approaches are difficult to obtain the complete edge information. Snake algorithm is a reasonable approach for edge detection. An improved initial contour Snake model is proposed in this paper. A rectangle is first located to surround the edge to be detected instead of drawing a series of points as initial contour. Then, we map these points on the rectangle to the edge surrounded according to certain rule to constitute the initial contour. After the mapping strategy, Snake algorithm is used to iterate the initial contour. Experiments show that complete pores edge information are achieved. Based on the edge pixels, the area of each pores can be calculated easily.