Extraction of Ground Photon for Photon Counting Lidar Over Rugged Mountainous Area Based on Wavelet Transform With Adaptive Window

With its high penetration, spaceborne photon-counting LiDAR can acquire the vertical structure of vegetation, which can be used for high-precision inversion of vegetation vertical parameters in forest remote sensing. Accurate distinguishing ground photons is the key to improving the inversion accuracy of forest structure parameters. In the current ground photon cloud extraction algorithms of spaceborne photon-counting LiDAR, the ground photon cloud is mostly extracted using a fixed window size, which is weak in rugged mountainous areas. In this letter, a window adaptive algorithm based on particle swarm optimization (PSO) is proposed to extract ground photon point clouds from de-noised photon point cloud data. Through adaptive optimization, the classification window is dynamically determined, which solves the inadaptability of the fixed window size change to the changing target characteristics, reduces the classification error, and also has strong adaptability in the rugged mountain forest area. Compared with the digital terrain model (DTM) data, the extraction accuracy of the proposed algorithm was significantly improved, with an optimal root mean square error (RMSE) of 0.95 m and a Bias of −0.27 m, indicating that the proposed algorithm could be well applied to the extraction of photon point clouds on the ground of rugged mountains.