The recognition of impact craters can provide information on impact craters' history and Martian evolution processes as well as for Martian rover landing site. Most previous studies about machine detection of impact craters on Mars are based on imagery data, however, it has large uncertainty in image processing. In this study, we present a novel approach for automatic recognition of impact craters based on the Mars Orbiter Laser Altimeter (MOLA)-derived digital topography data. Topographic curvature, which delineates impact crater on Digital Elevation Model (DEM), can be deduced from topography data. The thresholding map of curvature is transformed into a binary map, from which we can detect impact craters by combination of segmentation and flooding algorithms. More impact craters on Mars with confirmation algorithm can be effectively distinguished truly, which are added the existing catalog of manually identified Martian craters. It will be more useful for the study on impact craters' history and Martian evolution processes as well as Martian rover landing navigation.