基于DPC算法的星载激光雷达数据去噪方法在水深测量中的应用  

Application of DPC algorithm-based denoising method for satellite-borne lidar data to bathymetry

作　　者：陈鲁宾 李杰 孟文君 董志鹏 唐秋华 CHEN Lubin;LI Jie;MENG Wenjun;DONG Zhipeng;TANG Qiuhua(College of Geodesy and Geomatics,Shandong Untiversity of Science and Technology,Qingdao 266590,China;First Institute of Oceanography,Ministry of Natural Resources,Qingdao 266061,China;Key Laboratory of Oceanic Surveying and Mapping,Ministry of Natural Resources,Qingdao 266590,China)

机构地区：[1]山东科技大学测绘与空间信息学院,山东青岛266590 [2]自然资源部第一海洋研究所,山东青岛266061 [3]自然资源部海洋测绘重点实验室,山东青岛266590

出　　处：《海洋通报》2024年第5期632-638,共7页Marine Science Bulletin

基　　金：国家重点研发计划(2023YFC3107601);国家自然科学基金(41876111);山东省自然科学基金(ZR2023MD073);南极重点海域对气候变化的响应和影响(IRASCC2020-2022)。

摘　　要：美国冰、云和陆地高程二号卫星(The Ice,Cloud,and Land Elevation Satellite-2,ICESat-2)是ICESat卫星的继任者,旨在监测地球的冰盖、冰川、海洋和陆地高程的变化等,其携带的地形激光高度计系统(ATLAS)发射532 nm波长的激光,具备一定的水体穿透能力。作为光子计数式激光雷达,ICESat-2的数据易受外界环境影响而接收到大量噪声光子,导致光子数据密度分布不均匀。本文提出了一种基于密度峰值聚类(Density Peak Clustering,DPC)算法的光子去噪方法,通过数据集的欧式距离计算局部密度作为点云数据的属性,采用基尼指数自适应选择最优截断距离,分别对日间和夜间数据进行多次实验,得出了两类数据的局部密度阈值参数。本文选取三处实验区域进行信号光子去噪分析,使用本文方法的去噪精度F值优于官方置信度标签去噪和传统密度聚类算法(Density-Based Spatial Clustering of Applications with Noise,DBSCAN),可以应用于星载激光雷达数据去噪处理。最后,对去噪后的华光礁区域信号光子进行折射校正,与收集的DEM数据进行对比可见,结合本文去噪方法可以使用ICESat-2数据进行浅水域的水深测量。The U.S.Ice,Cloud,and Land Elevation Satellite-2(ICESat-2),succeeding the ICESat satellite,is designed to monitor changes in Earth's ice caps,glaciers,oceans,and land elevation.It carries the Advanced Topographic Laser Altimeter System(ATLAS)that emits a 532 nm laser that penetrates water bodies to a certain extent.As a photon-counting LiDAR system,ICESat-2's data is susceptible to external environmental factors,and receives a large number of noisy photons,which leads to uneven photon data distribution.This paper proposes a photon denoising method based on the Density Peak Clustering(DPC)algorithm,which calculates local density as a point cloud data attribute through the Euclidean distance within the dataset.The optimal truncation distance is adaptively selected using the Gini index,followed by experiments on daytime and nighttime data to derive threshold parameters for local density in both data types.Three experimental areas are selected for signal photon denoising analysis.The denoising accuracy F-value using this method is higher than official confidence label denoising and the denoising clustering algorithm(DBSCAN),making it suitable for satellite-borne LiDAR data denoising.Finally,the refraction correction of the signal photons in the denoised Huaguang Reef area was performed,and the comparison with the collected DEM data shows that the denoising method presented in this paper can effectively enable bathymetry measurements in shallow waters using ICESat-2 data.

关 键 词：ICESat-2卫星 光子去噪 水深测深 DPC算法 基尼指数 

分 类 号：P229.1[天文地球—大地测量学与测量工程]