一种用于机载LiDAR波形数据高斯分解的高斯拐点匹配法  

A Gaussian Inflexion Points Matching Method for Gaussian Decomposition of Airborne LiDAR Waveform Data

作　　者：段乙好[1,2] 张爱武[1,2] 刘诏 王书民[3] 王京萌[1,2] 叶秋虹 

机构地区：[1]首都师范大学三维信息获取与应用教育部重点实验室,北京100048 [2]首都师范大学空间信息技术教育部工程研究中心,北京100048 [3]中国地震局地震预测研究所,北京100036

出　　处：《激光与光电子学进展》2014年第10期189-197,共9页Laser & Optoelectronics Progress

基　　金：国家科技支撑计划(2012BAH31B01);北京市自然科学基金重点项目(KZ201310028035)

摘　　要：针对小光斑全波形机载激光雷达(LiDAR)波形数据高斯分解法的核心问题——高斯分量个数估计,提出一种高斯拐点匹配法。该算法用平面曲线离散点集拐点的快速查找算法检测波形数据中的拐点,计算过检测出的拐点及其右边第一个点的直线的斜率,根据斜率将所有检测出的拐点分为左、右拐点,一个左拐点与其邻近的一个右拐点组成一个高斯分量,据此可以确定波形数据中高斯分量个数。采用高斯拐点匹配法对模拟和实测波形数据进行分解,并与传统的脉冲检测方法(重心法和高斯脉冲拟合法)相比。结果表明,高斯拐点匹配法方法能极大地减小伪拐点的影响,快速、准确地检测并分解出波形数据中高斯分量,提高波形数据分解速度。同时其能分解出更多的高斯分量,从而提高点云密度。Estimation of Gaussion components＇ number is a core problem in the procedure of Gaussian decomposition of small-footprint full-waveform airborne LiDAR waveform data. A new approach named Gaussian inflexion points matching method （GIPM） is proposed to solve it. GIPM algorithm uses the quick locating algorithm for turning points in discrete point set of plane curve （QLATP） method for detecting the inflexion points （IFPs）. The slope of the line between the detected IFP and its adjacent point is calculated. The detected IFPs are classified as left IFPs and right IFPs according to the slope. A left IFP and its neighboring right IFP comprise a Gaussian component, thus getting the number of the Gaussian components of waveform data. GIPM method is used to decompose the simulated and the measured waveform data, comparing with two traditional pulses detection method （center of gravity and Gaussian pulse fitting）. The results demonstrate that the GIPM method can tremendously retain the impact of the pseudo IFPs, and quickly and accurately detect and decompose Gaussian components of the waveform data, and then immensely speed up the decomposition of waveform data. Meanwhile, it can get more Gaussian components than others, thus improving the density of point cloud.

关 键 词：遥感 机载激光雷达 波形分析 高斯分解 高斯拐点匹配法 全波形数据 

分 类 号：P208[天文地球—地图制图学与地理信息工程]