The study of impact craters is crucial for understanding planetary evolution and geological processes, particularly small craters, which are key to reconstructing the lunar impact history. Detecting small craters, with diameters ranging from 0.2 to 2 km, remains a challenge due to the power-law distribution of crater sizes and the complex topography of the lunar surface. This work uses high-resolution lunar imagery data from the Chang’E-2 mission, with a 7 m spatial resolution, to develop a deep learning framework for small crater detection, named YOLO-SCNet. The framework combines a high-quality, diversified sample dataset, generated through data augmentation techniques, with YOLO-SCNet, specifically designed for small target detection. Key challenges in lunar crater detection, such as varying lighting conditions and complex terrains, are addressed through the innovative model architecture, which incorporates a small object detection head, dynamic anchor boxes, and multi-scale feature fusion. Experimental results demonstrate that YOLO-SCNet achieves outstanding performance in detecting small craters across different lunar regions, with precision, recall, and F1 scores of 90.2%, 88.7%, and 89.4%, respectively. The framework offers a scalable solution for constructing a global lunar crater catalog (≥0.2 km) and can be extended to other planetary bodies like Mars and Mercury, significantly supporting future planetary exploration and mapping efforts.

Taking the Chang’e-4 and Chang’e-5 landing areas as the study areas, this study extracts the geological unit information from the regional USGS geological map, as well as the feature information such as topography and geomorphology, material composition and mineral abundance from Chang’e-2 DOM and DEM, wide angle camera (WAC) and Kaguya multi-band imager data. By applying methods including the statistical-based estimation of mutual information of data and the integrated-algorithmic-model-based evaluation of feature importance to this extracted information, we screen the significant features and construct a high-precision classification model by combining machine learning algorithm with important features of sample data. The practical application of the multi-classification prediction on the complex geological units in the two study areas achieves 97.9% and 95.1% accuracy. At the same time, the significant characteristics of the study area are mined, and the rules and knowledge associated with the geological evolution of the study area are obtained. In this study, we carry out research on quantitative prediction and identification of lunar surface geological units based on large samples and construct a high-precision multi-classification model to achieve automatic classification and prediction on large sample geological units with high accuracy. This method provides a new idea for the predicted mapping of geological units of lunar global digital mapping. In addition, it helps to fully exploit the useful information in the data and enrich the knowledge regarding the formation and evolution of the Moon.

The study of impact craters is crucial in understanding planetary evolution and geological processes, particularly small craters, which are essential for reconstructing the lunar impact history and geological timeline. However, due to the power-law distribution of crater sizes and the complex topography of the lunar surface, detecting small craters on a global scale remains a significant challenge. This study proposes an innovative sample creation method and develops a deep learning framework for small target detection, YOLO-SCNet, aimed at detecting small lunar craters with diameters ranging from 0.2 to 2 kilometers. By combining a high-quality, diversified sample dataset generated using data augmentation techniques with the YOLO-SCNet, specifically designed for small target detection, we successfully addressed key challenges in lunar crater detection. Experimental results show that YOLO-SCNet excels in adapting to complex terrains and varying lighting conditions, achieving outstanding performance in detecting small craters across different lunar regions, with precision, recall, and F1 scores of 90.2%, 88.7%, and 89.4%, respectively. The YOLO-SCNet framework demonstrates the immense potential of deep learning technologies in planetary geological research. It can be used to construct a global, high-precision lunar crater catalog (≥0.2 km), helping to fill the gap in the global lunar crater database for small craters. Moreover, the framework is highly scalable, with the potential to be extended to other planetary bodies, such as Mars and Mercury, providing significant support for future planetary exploration and mapping tasks.