Adaptive Denoising Algorithm Of Photon Counting LiDAR Point Cloud

Photon counting LiDAR point cloud denoising is the important and difficult points of ground data processing,which determines the accuracy and efficiency of subsequent processing.This paper focuses on the research of the photon counting LiDAR point cloud denoising algorithm,summarizes and compares the existing research results of traditional LiDAR and photon counting LiDAR denoising;Based on the systematic study of the photon counting LiDAR's related theories,combined with its data characteristics,an adaptive denoising algorithm is designed and a comparative experiment is carried out.The performance verification and targeted analysis of the denoising algorithm are carried out by using three types of data,such as The Ice,Cloud and Land Elevation Satellite-2.In summary,the main work and innovations completed by this paper are as follows:1.The significance of point cloud denoising for LiDAR data processing is explained,the domestic and international development of photon counting LiDAR payload is combed,the denoising algorithm of the existing traditional linear detection LiDAR point cloud and photon counting LiDAR point cloud is summarized and compared.2.The basic theory of photon counting LiDAR is summarized from detection principle,signal and noise model,detection probability and ranging principle;Focusing on the point cloud's data structure of The Ice,Cloud and Land Elevation Satellite-2,the storage format and indexing method of the photon counting LiDAR point cloud are researched and analyzed;The advantages and disadvantages of Poisson denoising,Canny edge detection denoising and spatial density denoising are compared;The terrain restoration algorithm is in-depth analyzed.3.An adaptive denoising algorithm for photon counting LiDAR point cloud is designed.The algorithm proposes an adaptive P-value selection strategy.The optimal P-value is sought by iterative approach through the dynamic evaluation of the point cloud to improve the applicability of the denoising kernel.The algorithm proposes a method that fits the signal and noise to a Gaussian function,and realizes the accurate separation of the signal and noise on the fitted waveform.At the same time,the EM algorithm is introduced into the Gaussian function fitting process to realize the denoising threshold's adaptive adjustment according to different payloads,different terrains,and different time periods.Aiming at some non-"typical" residual cluster noise of point clouds,the algorithm designed a secondary elevation histogram denoising to further improve the accuracy.4.Based on the three data of domestic airborne Multiple Beam LiDAR,foreign airborne Multiple Altimeter Beam Experimental LiDAR and foreign satellite-borne The Ice,Cloud and Land Elevation Satellite-2,the experiment of photon counting LiDAR point cloud denoising under the conditions of different terrains are developed,the adaptive denoising performance of different payload data is verified and analyzed,and the accuracy of the processing results is evaluated.The denoising experiment of The Ice,Cloud and Land Elevation Satellite-2 fills in the lack of research on the denoising algorithm of spaceborne photon counting LiDAR at home,and provides the foundation and support for the domestic research of the spaceborne photon counting LiDAR.