In order to solve the problem of excessive noise point clouds in the photon counting lidar detection data, a single photon point cloud denoising method based on a combination of improved density-based spatial clustering of applications with noise (DBSCAN) algorithm and statistical filtering algorithm was adopted. The actual flight data of multiple altimeter beam experimental lidar provided by National Aeronautics and Space Administration was experimental data. First, the point cloud density was obtained through the k-dimensional tree for rough denoising, and then the improved DBSCAN algorithm and statistical filtering algorithm were used for fine denoising. The theoretical analysis and experimental verification has achieved good results. The results show that the target point cloud recognition rate in the experimental area is above 85%, and the performance is better than the classic radius filtering algorithm. This result is helpful for photon data denoising.