The second-generation laser altimetry satellite ICESat-2, launched by the United States on September 15, 2018, uses micropulse photon counting lidar technology for the first time and performs elevation profiling type Earth observation. It is capable of realizing the global acquisition of geomorphological information with fast and efficient characteristics. Its advantages of micropulses, high frequency repetition, multiple beams and small footprint have made it a trend in the development of active earth observation laser altimetry in the future. Therefore, research on the processing and application methods of point cloud data collected by ICESat-2 is crucial. In this paper, the single photon laser point cloud data collected by ICESat-2 / ATLAS is taken as an example. Firstly, the OPTICS clustering algorithm is used to denoise the point cloud data, and then the least squares curve fitting method is used to classify the ground points and non-ground points by using the difference between the theoretical value and the actual value of the ground point elevation. Set a reasonable threshold to classify the ground points and non-ground points. The experiments in Shanghai and Qingdao prove the effectiveness of the two methods. The accuracy of denoising and classification can reach more than 95%, which can be applied to the processing of single-photon laser point cloud data.