In this paper,a methodology for automatically extracting impact craters from planet surface based on image data is presented according to that the sunny side of crater is bright and the back side is dark. In the instance,the optical image from the planet surface is segmented by using the adaptive double threshold segmentation method,and both shape and position information of light and shade area in the image are obtained. Then,the structures of light and shade areas are analyzed according to the principle of statistics,and the direction of illumination is estimated in image-plane coordinate system. Combined with correlated constraint conditions,light and shade areas from same crater are matched,the external edge of the impact crater is fitted,and at the same time,the information of radius and position are obtained. The experiment results show that the method can be used to quickly and reliably extract both center position and radius of impact craters from planetary optical image,and it has wide adaptability and some certain application value.