收藏设为首页 广告服务联系我们在线留言

干旱区地理 ›› 2017, Vol. 40 ›› Issue (6): 1256-1263.

• 地球信息科学 • 上一篇    下一篇

近地机载激光雷达棉花LAI提取方法研究

陈洪1,2,3,4, 韩峰1,2,3,4, 赵庆展1,2,3,4, 刘伟1,2,3,4, 张天毅2,3,4   

  1. 1. 石河子大学信息科学与技术学院, 新疆 石河子 832000;
    2. 国家遥感中心新疆兵团分部, 新疆 石河子 832000;
    3. 兵团空间信息工程技术研究中心, 新疆 石河子 832000;
    4. 兵团空间信息工程实验室, 新疆 石河子 832000
  • 收稿日期:2017-07-11 修回日期:2017-09-21 出版日期:2017-11-25
  • 通讯作者: 韩峰(1979-),男,讲师,主要从事数字图像处理与机器视觉方向研究.E-mail:hf_inf@shzu.edu.cn
  • 作者简介:陈洪(1991-),男,硕士,主要从事空间信息技术及应用、无人机遥感的应用基础研究.E-mail:714772559@qq.com
  • 基金资助:

    新疆生产建设兵团科技计划项目(2016BA001);中央引导地方科技发展专项资金项目(BT011)

One extraction method of cotton LAI from airborne Li-DAR of UAV

CHEN Hong1,2,3,4, HAN Feng1,2,3,4, ZHAO Qing-zhan1,2,3,4, LIU Wei1,2,3,4, ZHANG Tianyi2,3,4   

  1. 1. College of Information Science and Technology, Shihezi University, Shihezi 832000, Xinjiang, China;
    2. Xinjiang Production and Construction Corps Division, National Remote Sensing Center of China, Shihezi 832000, Xinjiang, China;
    3. Geospatial Information Engineering Research Center, Xinjiang Production and Construction Corps, Shihezi 832000, Xinjiang, China;
    4. Laboratory of Geospatial Information Engineering, Xinjiang Production and Construction Corps, Shihezi 832000, Xinjiang, China
  • Received:2017-07-11 Revised:2017-09-21 Online:2017-11-25

摘要: 棉花叶面积指数(LAI)是描述其长势的重要指标,准确获取冠层结构参数是叶面积指数反演的必要条件。以ScoutB-100油动单旋翼无人机为飞行平台,搭载RIEGL VUX-1激光雷达,精确获取棉花高密度点云数据,得到研究区棉田数字表面模型(DSM)和数字高程模型(DEM),通过差值运算获得其冠层高度模型(CHM),进而提取有效的冠层结构参数。利用相关性分析法选取相关系数大于0.2的激光穿透力指数(LPI)、回波点云密度(D)、孔隙率(fgap)、归一化高程值(VnDSM)构建棉花LAI反演模型,并与实测叶面积指数进行精度验证与评价。实验结果表明:模型估算的LAI与实测LAI之间的决定系数为0.824,均方根误差为0.072,验证了模型的可靠性。

关键词: 机载激光雷达, 冠层高度模型, 棉花, 叶面积指数

Abstract: The leaf area index (LAI)plays an important role for describing the growth of the cotton. It is necessary to obtain the canopy structure parameters accurately for the inversion of leaf area index. And the light detection and ranging (LiDAR)provide a way to obtain the parameters accurately. So far,domestic and international studies about LAI are mainly focused on forest. The researches on extracting the crop parameters with point cloud data are still in the initial stage,and the method of using laser scanner to retrieve the leaf area index of crop needs further verification. On the other hand,there are little researches on extracting the crop parameters with low density of point cloud data. How about using high density of point cloud data to extract the crop parameters? In this paper, the researchers obtain high-density point cloud data from farmland by keeping the unmanned aerial vehicle (UAV)flying low,and in this way,the relationship between the vegetation parameters and the LAI is explored, the point cloud density is 299 pts·m-2. By using the Scout B-100 oil-powered single-rotor UAV as the flight platform, RIEGL VUX-1 obtained high accurate point cloud data. Firstly,using the laser scanner to get point cloud data,the density of which is 299 pts·m-2 from Mushroom Lake Regional Farmland in northwest of Shihezi City, and the point cloud filtering is used to separate the ground and the vegetation;Secondly,the digital surface model (DSM)and digital elevation model (DEM)of the study area were affected by point cloud data. Thirdly,the canopy height model (CHM)was obtained by calculating the difference between DSM and DEM,after which,the effective canopy structure parameters were extracted from the CHM by the watershed algorithm. In order to construct LAI cotton inversion model,correlation analysis,selecting the correlation coefficient,which is greater than 0.2,including laser penetration index (LPI),echo point cloud density (D),gap fraction (fgap)and value of normalized elevation (VnDSM)has been made. Finally,verify the accuracy and evaluation with the measured leaf area index. The experimental results show that R2 between the estimated LAI and the measured LAI is 0.824,and the RMSE is 0.072,which verifies the reliability of the model. The filtering method of point requires further investigation, and the method of collecting high-density point cloud data in the low vegetation area is still not explicit. Considering the small scale of the cotton,the research does not choose the leaf angle as the model input parameters, and the results are satisfactory. Compared with other crops LAI parameter inversion method,it is more efficient and faster than the others. By showing that using high density of point cloud data to extract the cotton leaf area index in the study area,it is concluded that this method can also be applied to point cloud data with high precision satisfactorily.

Key words: airborne LiDAR, canopy height model, cotton, LAI

中图分类号: 

  • TP79