大光斑LiDAR全波形数据小波变换的高斯递进分解
Received:April 14, 2017  Revised:September 14, 2017  点此下载全文
引用本文:杨学博,王 成,习晓环,田建林,聂 胜,朱笑笑.大光斑LiDAR全波形数据小波变换的高斯递进分解[J].Journal of Infrared and Millimeter Waves,2017,36(6):749~755
Hits: 66
Download times: 306
Author NameAffiliationE-mail
YANG Xue-Bo Key Laboratory of Digital Earth ScienceInstitute of Remote Sensing and Digital Earth, Chinese Academy of Sciences yangxb@radi.ac.cn 
WANG Cheng Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences wangcheng@radi.ac.cn 
XI Xiao-Huan Institute of Remote Sensing and Digital Earth,Chinese Academy of Sciences xhxi@ceode.ac.cn 
TIAN Jian-Lin Sun Yat-sen University 185363236@qq.com 
NIE Sheng Institute of Remote Sensing and Digital Earth,Chinese Academy of Sciences niesheng@radi.ac.cn 
ZHU Xiao-Xiao Institute of Remote Sensing and Digital Earth,Chinese Academy of Sciences zhuxx@radi.ac.cn 
基金项目:国家重点基础研究发展计划(973计划)(2017YFA0603002),国家自然科学基金面上项目(41271428)
中文摘要:高斯分解是波形激光雷达数据预处理的常用方法,但在应用于大光斑全波形激光雷达数据中的叠加波时却难以发挥作用,为此提出一种基于小波变换的高斯递进波形分解方法。首先,利用小波变换多尺度分析特性检测出目标地物并准确估算组分特征参数,进而建立高斯模型优化特征参数;然后通过拟合精度指标,判断是否需要添加新组分进行逐级递进分解,确定最终模型及其组分构成,最终实现全波形激光雷达数据的波形分解。为了验证算法的有效性,分别对实验数据使用本文算法和常用的基于拐点匹配的高斯分解法进行分析,结果表明,本文算法提取的目标数几乎是拐点匹配算法的2倍,可以有效地从叠加波中检测出目标组分,且拟合精度高于98%。
中文关键词:大光斑激光雷达  全波形分析  小波变换  高斯分解  特征参数
 
Wavelet transform of Gaussian progressive decomposition method for full-waveform LiDAR data
Abstract:Gaussian decomposition is the most commonly used methods for waveform analysis, which is a key post-processing step for the applications of space-borne LiDAR data. However, it usually fails to detect the overlapping pulses of large-footprint waveform data. Therefore, a Gaussian progressive decomposition method based on wavelet transform was proposed in this study to address this issue and applied to Ice, Cloud, and land Elevation Satellite / Geoscience Laser Altimeter System (ICESat/GLAS) data. The new proposed method mainly consists of three key steps. First, the wavelet transform was adopted to detect the target features and estimate the component feature parameters, then the Gaussian model was established to optimize the feature parameters. Second, a new component was added if the fitting accuracy didn’t meet the requirements. Finally, waveform decomposition based on wavelet transform was completed until no more new components were added. Additionally, a comparison experiment between the new proposed method and the Gaussian decomposition method based on inflection point was also conducted to verify the reliability of the new proposed algorithm. Experiment results indicated that our new proposed algorithm can detect twice targets as many as the method based on inflection point, and effectively decompose the targets from overlapping waveforms due to high fitting accuracy of above 98%.
keywords:Large footprint LiDAR  full-waveform analysis  wavelet transform  Gaussian decomposition  characteristic parameter
View Full Text  View/Add Comment  Download reader

Copyright:《Journal of Infrared And Millimeter Waves》