Single photon point cloud denoising algorithm based on multi-features adaptive

The new spaceborne photon counting radar can acquire high-precision three-dimensional information of ground and ground targets, but its measurement accuracy is greatly affected by noise. Aiming at the difficulty of signal extraction of single-photon laser data in areas with inconsistent background noise and large slope area, this paper proposed a single photon point cloud denoising algorithm based on multi-feature adaptive. It was different from the traditional circular or elliptical filtering kernel, and used the parallelogram filtering kernel which was more in line with the characteristics of single photon point cloud data, and signals were adaptively identified by slope, spatial density and noise rate. The ICESat-2 single photon point cloud data located in the glacier area of Qinghai-Tibet Plateau was selected to carry out the point cloud denoising test and verification, and the study area had a large slope and broken terrain. Compared with the official denoising results of ATL03 and ATL08, the proposed algorithm has better performance in areas with inconsistent background noise level and large slope area.