With the continuous development of deep space exploration technology and planetary scientific research, the scientific tasks and technical breakthroughs undertaken by China in the field of lunar and small celestial body exploration have become increasingly important. However, previous studies on the identification and quantitative research of craters (or impact craters) were limited by factors such as data scale and traditional algorithms, making it difficult to meet the requirements for crater identification in complex terrain conditions. Therefore, to solve these problems, this study employs: deep convolutional neural networks (CNN), high-resolution digital elevation models (DEM), etc., to achieve the automatic identification and edge extraction of craters, improving the identification ability and enhancing the scientific credibility of the model. We expect these solutions to provide technical support for future lunar base site selection, asteroid defense, and exploration of extraterrestrial life.