Volume 39 Issue 4
Jul.  2022
Turn off MathJax
Article Contents
Abstract
References
Article Navigation >  Journal of Zhejiang A&F University  > 2022  >  39(4): 800-806

HUANG Haiyan, TANG Huimin, JIN Lu, et al. Effects of tree species importance, diversity and soil physicochemical properties on soil amino sugars[J]. Journal of Zhejiang A&F University, 2024, 41(4): 778-786. DOI: 10.11833/j.issn.2095-0756.20230527
Citation: CHEN Siyu, LIU Xianzhao, WANG Yixiang, et al. Individual tree detection in the high canopy density Larix principis-rupprechtii forest based on airborne LiDAR[J]. Journal of Zhejiang A&F University, 2022, 39(4): 800-806. DOI: 10.11833/j.issn.2095-0756.20210399
shu

Individual tree detection in the high canopy density Larix principis-rupprechtii forest based on airborne LiDAR

DOI: 10.11833/j.issn.2095-0756.20210399
  • 1.

    College of Environment and Resources, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China

  • 2.

    State Key Laboratory of Subtropical Silviculture, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China

  • 3.

    Zhejiang Provincial Key Laboratory of Carbon Cycling in Forest Ecosystems and Carbon Sequestration, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China

  • 4.

    Key Laboratory of Forest Management and Growth Modeling, National Forestry and Grassland Administration, Research Institute of Forest Resources Information Techniques, Chinese Academy of Forestry, Beijing 100091, China

  • Received Date: 2021-05-31
  • Accepted Date: 2022-03-25
  • Rev Recd Date: 2022-03-25
    • Available Online: 2022-07-20
  • Publish Date: 2022-08-20
  •   Objective  With low identification accuracy of individual trees in Larix principis-rupprechtii forest with high canopy density employing high resolution images, this paper is aimed to confirm the strengths of airborne Laser Detection and Ranging (LiDAR) 3D point cloud data as an alternative with a workable method proposed.   Method  Based on the preprocessing of point cloud data, an improved Mean Shift with Gaussian kernel function (MSP) position recognition method on the basis of the spatial characteristics of airborne LiDAR point cloud was proposed. The comparison is made with other three commonly used methods: regional growing segmentation algorithm based on point cloud (RGP), local maximum method based on canopy height model (LMC) and multi-scale segmentation method based on CHM (MSC).   Result  Identification accuracy of the four methods is: MSP (89.30%)＞LMC (85.60%)＞RGP (77.50%)＞MSC (70.00%), and MSC, the proposed method, displayed high average individual tree crown extraction accuracy (90.18%) and relatively low omission error and commission error rate: 8.7% and 8.0% respectively.   Conclusion  The proposed MSP has good applicability in high crown density L. principis-rupprechtii forest and provides a new way of extracting L. principis-rupprechtii forest structure parameters accurately on the basis of airborne LiDAR point clouds. [Ch, 3 fig. 3 tab. 28 ref.]
  • [1] HAN Xinsheng, WANG Xiao, WANG Yanhui, LI Zhenhua.  Difference of daily understory evapotranspiration under two water conditions in Larix principis-rupprechtii plantation . Journal of Zhejiang A&F University, 2024, 41(5): 949-958. doi: 10.11833/j.issn.2095-0756.20240251
    [2] WANG Yuqi, ZHANG Qianqian, ZHANG Wenzhuo, YU Yefei, LÜ Qiangfeng, TENG Qiumei, LI Yongchun.  Response of yield and tetrahydropalmatine content of Corydalis yanhusuo to canopy density and litter treatment of Chinese fir forest . Journal of Zhejiang A&F University, 2024, 42(): 1-11. doi: 10.11833/j.issn.2095-0756.20240461
    [3] XU Shanshan, LI Changchun, ZHANG Chao.  Extraction of DBH and height of Pinus yunnanensis individual trees by combining backpack laser scanning and unmanned aerial vehicle laser scanning . Journal of Zhejiang A&F University, 2024, 41(5): 939-948. doi: 10.11833/j.issn.2095-0756.20240107
    [4] YU Bao.  Characteristics and impact factors of self-pruning in natural Larix gmelinii forest . Journal of Zhejiang A&F University, 2023, 40(1): 209-216. doi: 10.11833/j.issn.2095-0756.20220220
    [5] WANG Yuyang, WANG Yixiang, LI Mingzhe, LIANG Dan.  Estimation method of Phyllostachys edulis forest canopy density based on UAV visible image . Journal of Zhejiang A&F University, 2022, 39(5): 981-988. doi: 10.11833/j.issn.2095-0756.20210576
    [6] WU Xiujuan, AO Xiaoping, ZHAO Yupeng, CUI Xueqing.  Spatial structure of Larix principis-rupprechtii-Picea spp. secondary forests on shady slope of Luyashan National Nature Reserve . Journal of Zhejiang A&F University, 2021, 38(1): 58-64. doi: 10.11833/j.issn.2095-0756.20200261
    [7] YU Bao.  Characteristics and impact factors of space utilization of young and middle-aged natural Larix gmelinii forests . Journal of Zhejiang A&F University, 2020, 37(3): 407-415. doi: 10.11833/j.issn.2095-0756.20190382
    [8] SHEN Jiapeng, CHEN Dongsheng, SUN Xiaomei, ZHANG Shougong.  Modeling a single-tree biomass equation by seemingly unrelated regression and dummy variables with Larix kaempferi . Journal of Zhejiang A&F University, 2019, 36(5): 877-885. doi: 10.11833/j.issn.2095-0756.2019.05.005
    [9] TAO Jiangyue, LIU Lijuan, PANG Yong, LI Dengqiu, FENG Yunyun, WANG Xue, DING Youli, PENG Qiong, XIAO Wenhui.  Automatic identification of tree species based on airborne LiDAR and hyperspectral data . Journal of Zhejiang A&F University, 2018, 35(2): 314-323. doi: 10.11833/j.issn.2095-0756.2018.02.016
    [10] FENG Yunyun, LIU Lijuan, LU Dengsheng, PANG Yong.  A comparative study of dimension reduction methods for airborne hyperspectral images . Journal of Zhejiang A&F University, 2017, 34(5): 765-774. doi: 10.11833/j.issn.2095-0756.2017.05.001
    [11] ZHAO Kuangji, JI Fuli, LIU Yanwen, LIU Xiaolan, JIA Zhongkui, MA Lüyi.  Growth of Larix principis-rupprechtii with thinning and pruning . Journal of Zhejiang A&F University, 2016, 33(4): 581-588. doi: 10.11833/j.issn.2095-0756.2016.04.005
    [12] SUN Pengyue, XU Fuli, WANG Weiling, WANG Lingling, NIU Ruilong, GAO Xing, BAI Xiaofang.  Seasonal dynamics of soil nutrients and soil enzyme activities in Larix principis-rupprechtii plantations . Journal of Zhejiang A&F University, 2016, 33(6): 944-952. doi: 10.11833/j.issn.2095-0756.2016.06.004
    [13] YAO Zhi, ZHANG Xiaoli.  Based on WebGIS platform developing remote sensing information systems of forest canopy density inversion . Journal of Zhejiang A&F University, 2015, 32(3): 392-398. doi: 10.11833/j.issn.2095-0756.2015.03.009
    [14] ZHAO Xiao, LÜ Yulong, WANG Cong, LI Yadan, DU Huaqiang.  Spatial distribution for leaf area index and canopy closure of Phyllostachys edulis stand using Cokriging . Journal of Zhejiang A&F University, 2014, 31(4): 560-569. doi: 10.11833/j.issn.2095-0756.2014.04.011
    [15] ZHANG Zhen, ZHANG Han-guo, ZHANG Lei, ZHU Hang-yong, LI Xue-feng.  Diversity analysis with RAPD for natural and breeding populations of Larix gmelinii . Journal of Zhejiang A&F University, 2012, 29(1): 130-136. doi: 10.11833/j.issn.2095-0756.2012.01.022
    [16] SUN Yu, LI Guo-lei, LIU Yong, MA Lü-yi, ZHU Yan, JIANG Chang-ji, LIU Fu-sen, LI Xue-lian.  Growth and P-uptake for Larix olgensis seedlings with phosphorus top-dressed using fertigation . Journal of Zhejiang A&F University, 2011, 28(2): 219-226. doi: 10.11833/j.issn.2095-0756.2011.02.008
    [17] YU Bao, ZHANG Qiu-liang, WANG Li-ming, WU Jisiguleng.  Characteristics of biomass and productivity in Larix gmelinii natural forests with different stand structures . Journal of Zhejiang A&F University, 2011, 28(1): 52-58. doi: 10.11833/j.issn.2095-0756.2011.01.009
    [18] ZHANG Lei, ZHANG Han-guo, DENG Ji-feng, GUAN Chun-yu.  Stability of hybrid larches （Larix） with seedling height growth . Journal of Zhejiang A&F University, 2010, 27(5): 706-712. doi: 10.11833/j.issn.2095-0756.2010.05.011
    [19] ZHANG Hui-ru, WU Ji-cheng, YANG Hong-bo, CHEN Xin-mei.  Spatial structure of mixed larch-spruce-fir stands . Journal of Zhejiang A&F University, 2009, 26(3): 319-325.
    [20] He Fuji, Wu Mingan, Ni Rongxin, Xie Zhengcheng, Zhang Jianzhong..  Effect of Canopy Density on Seed Yield of Chinese Fir in Seed Orchard. . Journal of Zhejiang A&F University, 1995, 12(3): 311-315.
  • [1]
    ZHAO Kuangji, WANG Lidong, WANG Lijun, et al. Stock volume and productivity of Larix principis-rupprechtii in northern and northwestern China [J]. J Beijing For Univ, 2015, 37(2): 24 − 31.
    [2]
    HILL S, LATIFI H, HEURICH M, et al. Individual-tree and stand-based development following natural disturbance in a heterogeneously structured forest: a LiDAR-based approach [J]. Ecol Inf, 2017, 38: 12 − 25.
    [3]
    LI Zengyuan, LIU Qingwang, PANG Yong. Review on forest parameters inversion using LiDAR [J]. J Remote Sensing, 2016, 20(5): 1138 − 1150.
    [4]
    SHEN Jiapeng, CHEN Dongsheng, SUN Xiaomei, et al. Modeling a single-tree biomass equation by seemingly unrelated regression and dummy variables with Larix kaempferi [J]. J Zhejiang A&F Univ, 2019, 36(5): 877 − 885.
    [5]
    ZHEN Zhen, LI Xiang, XIU Siyu, et al. Individual tree crown delineation using maker-controlled region growing method [J]. J Northeast For Univ, 2016, 44(10): 22 − 29.
    [6]
    TAO Jiangyue, LIU Lijuan, PANG Yong, et al. Automatic identification of tree species based on airborne LiDAR and hyperspectral data [J]. J Zhejiang A&F Univ, 2018, 35(2): 314 − 323.
    [7]
    HOLMGREN J, NILSSON M, OLSSON H. Estimation of tree height and stem volume on plots using airborne laser scanning [J]. For Sci, 2003, 49(3): 419 − 428.
    [8]
    HUO Da, XING Yanqiu, TIAN Xin, et al. Estimating individual tree crown diameter using fourth fegree polynomial fitting method based on airborne LiDAR [J]. J Northwest For Univ, 2015, 30(3): 164 − 169.
    [9]
    CHEN Wei, XIANG Haibing, MORIYA K. Individual tree position extraction and structural parameter retrieval based on airborne LiDAR data: performance evaluation and comparison of four algorithms [J/OL]. Remote Sensing, 2019, 12: 571[2021- 05-30]. doi: 10.3390/rs12030571.
    [10]
    WANG Pu, XING Yanqiu, WANG Cheng, et al. A graph cut-based approach for individual tree detection using airborne LiDAR data [J]. J Univ Chin Acad Sci, 2019, 36(3): 385 − 391.
    [11]
    LI Xiang, ZHEN Zhen, ZHAO Yinghui. Suitable model of detecting the position of individual treetop based on local maximum method [J]. J Beijing For Univ, 2015, 37(3): 27 − 33.
    [12]
    CHEN Shiyue, LIANG Dan, YING Binbin, et al. Assessment of an improved individual tree detection method based on local-maximum algorithm from unmanned aerial vehicle RGB imagery in overlapping canopy mountain forests [J]. Int J Remote Sensing, 2021, 42(1): 106 − 125.
    [13]
    CHO M A, MATHIEU R, ASNER G P, et al. Mapping tree species composition in South African savannas using an integrated airborne spectral and LiDAR system [J]. Remote Sensing Environ, 2012, 125: 214 − 226.
    [14]
    LIU Xiaoshuang, HUANG Jianwen, JU Hongbo. Research progress in the methods and application of individual tree crown’s automatic extracting by high spatial resolution remote sensing [J]. J Zhejiang For Coll, 2010, 27(1): 126 − 133.
    [15]
    LI Wenkai, GUO Qinghua, JAKUBOWSKI M K, et al. A new method for segmenting individual trees from the lidar point cloud [J]. Photogrammetric Eng Remote Sensing, 2012, 78(1): 75 − 84.
    [16]
    LI Senlei, LI Jianping, JIANG Tengping, et al. A multi-scale method for 3D individual tree extraction using UAV-LiDAR [J]. Geomatics Sci Technol, 2021, 9(1): 14 − 25.
    [17]
    NAVEED F, HU Baoxin, WANG Jianguo, et al. Individual tree crown delineation using multispectral LiDAR data [J/OL]. Sensors, 2019, 19: 5421[2021-05-30]. doi: 10.3390/s19245421.
    [18]
    LI Renzhong, LIU Yangyang, YANG Man, et al. Three-dimensional point cloud segmentation algorithm based on improved region growth [J/OL]. Laser Optoelectronics Prog, 2018, 55: 051502[2021-05-30]. doi: 10.3788/LOP55.051502.
    [19]
    KWONG I H Y, FUNG T. Tree height mapping and crown delineation using LiDAR, large format aerial photographs, and unmanned aerial vehicle photogrammetry in subtropical urban forest [J]. Int J Remote Sensing, 2020, 41(14): 5228 − 5256.
    [20]
    MA Zhengyu, PANG Yong, WANG Di, et al. Individual tree crown segmentation of a larch plantation using airborne laser scanning data based on region growing and canopy morphology features [J/OL]. Remote Sensing, 2020, 12(7): 1078[2021-05-30]. doi: 10.3389/rs12071078.
    [21]
    HU Xingbo, CHEN Wei, XU Weiyang. Adaptive mean shift-based identification of individual trees using airborne LiDAR data [J/OL]. Remote Sensing, 2017, 9(2): 148[2021-05-30]. doi: 10.3389/rs9020148.
    [22]
    CHEN Wei, HU Xingbo, CHEN Wen, et al. Airborne LiDAR remote sensing for individual tree forest inventory using trunk detection-aided mean shift clustering techniques [J/OL]. Remote Sensing, 2018, 10: 1078[2021-05-30]. doi: 10.3389/rs10071078.
    [23]
    YAN Wanqian, GUAN Haiyan, CAO Lin, et al. A self-adaptive mean shift tree-segmentation method using UAV LiDAR data [J/OL]. Remote Sensing, 2020, 12: 515[2021-05-30]. doi: 10.3389/rs12030515.
    [24]
    COMANICIU D, MEER P. Mean shift: a robust approach toward feature space analysis [J]. IEEE Trans Pattern Anal Mach Intel, 2002, 24(5): 603 − 619.
    [25]
    ZHANG Yizhuo, LÜ Akang, JIANG Dapeng, et al. Single tree segmentation and extraction of tree height and crown width using gaussian clustering [J]. J Northeast For Univ, 2021, 49(2): 54 − 59.
    [26]
    ZHANG Xueliang, XIAO Pengfeng, FENG Xuezhi. Multi-scale fast segmentation of remotely sensed image based on improved region adjacency graph [J]. Remote Sensing Inf, 2011(5): 3 − 8, 46.
    [27]
    JING Linhai, HU Baoxin, LI Jili, et al. Automated delineation of individual tree crowns from LiDAR data by multi-scale analysis and segmentation [J]. Photogramm Eng Remote Sensing, 2012, 78(12): 1275 − 1284.
    [28]
    HE Qisheng, CHEN Erxue, CAO Chunxiang, et al. A study of forest parameters mapping technique using airborme LiDAR data [J]. Adv Earth Sci, 2009, 24(7): 748 − 755.
  • Created with Highcharts 5.0.7Amount of accessChart context menuAbstract Views, HTML Views, PDF Downloads StatisticsAbstract ViewsHTML ViewsPDF Downloads2024-052024-062024-072024-082024-092024-102024-112024-122025-012025-022025-032025-040255075100Highcharts.com
    Created with Highcharts 5.0.7Chart context menuAccess Class DistributionFULLTEXT: 14.4 %FULLTEXT: 14.4 %META: 81.6 %META: 81.6 %PDF: 4.0 %PDF: 4.0 %FULLTEXTMETAPDFHighcharts.com
    Created with Highcharts 5.0.7Chart context menuAccess Area Distribution其他: 19.9 %其他: 19.9 %其他: 0.6 %其他: 0.6 %Beauharnois: 0.2 %Beauharnois: 0.2 %Rochester: 0.2 %Rochester: 0.2 %Seattle: 0.2 %Seattle: 0.2 %上海: 0.5 %上海: 0.5 %东莞: 0.2 %东莞: 0.2 %亳州: 0.3 %亳州: 0.3 %信阳: 0.2 %信阳: 0.2 %北京: 2.1 %北京: 2.1 %匹兹堡: 0.2 %匹兹堡: 0.2 %十堰: 1.0 %十堰: 1.0 %南京: 0.3 %南京: 0.3 %南宁: 0.5 %南宁: 0.5 %南昌: 0.5 %南昌: 0.5 %博阿努瓦: 0.6 %博阿努瓦: 0.6 %合肥: 0.5 %合肥: 0.5 %呼和浩特: 0.5 %呼和浩特: 0.5 %咸阳: 0.6 %咸阳: 0.6 %哈尔滨: 1.1 %哈尔滨: 1.1 %哥伦布: 0.2 %哥伦布: 0.2 %嘉兴: 0.5 %嘉兴: 0.5 %嘉峪关: 0.5 %嘉峪关: 0.5 %大急流城: 0.2 %大急流城: 0.2 %天津: 1.4 %天津: 1.4 %太原: 0.5 %太原: 0.5 %宣城: 0.6 %宣城: 0.6 %巴库: 0.2 %巴库: 0.2 %常德: 0.2 %常德: 0.2 %广州: 0.2 %广州: 0.2 %延安: 0.2 %延安: 0.2 %张家口: 2.6 %张家口: 2.6 %德里: 0.6 %德里: 0.6 %成都: 0.2 %成都: 0.2 %扬州: 2.6 %扬州: 2.6 %揭阳: 0.2 %揭阳: 0.2 %昆明: 0.3 %昆明: 0.3 %杭州: 3.2 %杭州: 3.2 %格兰特县: 0.3 %格兰特县: 0.3 %沈阳: 0.2 %沈阳: 0.2 %泉州: 0.5 %泉州: 0.5 %洛阳: 0.2 %洛阳: 0.2 %淄博: 0.2 %淄博: 0.2 %深圳: 0.6 %深圳: 0.6 %温州: 1.6 %温州: 1.6 %漯河: 5.6 %漯河: 5.6 %焦作: 1.0 %焦作: 1.0 %秦皇岛: 0.2 %秦皇岛: 0.2 %纽约: 0.2 %纽约: 0.2 %绍兴: 0.6 %绍兴: 0.6 %绵阳: 0.2 %绵阳: 0.2 %芒廷维尤: 16.0 %芒廷维尤: 16.0 %芝加哥: 1.3 %芝加哥: 1.3 %苏州: 1.0 %苏州: 1.0 %莆田: 0.2 %莆田: 0.2 %萨默维尔: 0.6 %萨默维尔: 0.6 %衡阳: 0.2 %衡阳: 0.2 %西宁: 16.3 %西宁: 16.3 %西安: 0.2 %西安: 0.2 %贵阳: 0.2 %贵阳: 0.2 %赫尔辛基: 0.3 %赫尔辛基: 0.3 %运城: 1.0 %运城: 1.0 %遵义: 0.3 %遵义: 0.3 %邯郸: 0.3 %邯郸: 0.3 %郑州: 2.4 %郑州: 2.4 %锦州: 0.5 %锦州: 0.5 %长春: 0.5 %长春: 0.5 %长沙: 3.2 %长沙: 3.2 %青岛: 0.3 %青岛: 0.3 %其他其他BeauharnoisRochesterSeattle上海东莞亳州信阳北京匹兹堡十堰南京南宁南昌博阿努瓦合肥呼和浩特咸阳哈尔滨哥伦布嘉兴嘉峪关大急流城天津太原宣城巴库常德广州延安张家口德里成都扬州揭阳昆明杭州格兰特县沈阳泉州洛阳淄博深圳温州漯河焦作秦皇岛纽约绍兴绵阳芒廷维尤芝加哥苏州莆田萨默维尔衡阳西宁西安贵阳赫尔辛基运城遵义邯郸郑州锦州长春长沙青岛Highcharts.com
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(3)  / Tables(3)

Get Citation
PDF
XML

Article views(786) PDF downloads(54) Cited by()

Related
Proportional views

Individual tree detection in the high canopy density Larix principis-rupprechtii forest based on airborne LiDAR

doi: 10.11833/j.issn.2095-0756.20210399
  • 1. College of Environment and Resources, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China
  • 2. State Key Laboratory of Subtropical Silviculture, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China
  • 3. Zhejiang Provincial Key Laboratory of Carbon Cycling in Forest Ecosystems and Carbon Sequestration, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China
  • 4. Key Laboratory of Forest Management and Growth Modeling, National Forestry and Grassland Administration, Research Institute of Forest Resources Information Techniques, Chinese Academy of Forestry, Beijing 100091, China
  • Received Date: 2021-05-31
  • Accepted Date: 2022-03-25
  • Rev Recd Date: 2022-03-25
  • Available Online: 2022-07-20
  • Publish Date: 2022-08-20

Abstract:   Objective  With low identification accuracy of individual trees in Larix principis-rupprechtii forest with high canopy density employing high resolution images, this paper is aimed to confirm the strengths of airborne Laser Detection and Ranging (LiDAR) 3D point cloud data as an alternative with a workable method proposed.   Method  Based on the preprocessing of point cloud data, an improved Mean Shift with Gaussian kernel function (MSP) position recognition method on the basis of the spatial characteristics of airborne LiDAR point cloud was proposed. The comparison is made with other three commonly used methods: regional growing segmentation algorithm based on point cloud (RGP), local maximum method based on canopy height model (LMC) and multi-scale segmentation method based on CHM (MSC).   Result  Identification accuracy of the four methods is: MSP (89.30%)＞LMC (85.60%)＞RGP (77.50%)＞MSC (70.00%), and MSC, the proposed method, displayed high average individual tree crown extraction accuracy (90.18%) and relatively low omission error and commission error rate: 8.7% and 8.0% respectively.   Conclusion  The proposed MSP has good applicability in high crown density L. principis-rupprechtii forest and provides a new way of extracting L. principis-rupprechtii forest structure parameters accurately on the basis of airborne LiDAR point clouds. [Ch, 3 fig. 3 tab. 28 ref.]

HUANG Haiyan, TANG Huimin, JIN Lu, et al. Effects of tree species importance, diversity and soil physicochemical properties on soil amino sugars[J]. Journal of Zhejiang A&F University, 2024, 41(4): 778-786. DOI: 10.11833/j.issn.2095-0756.20230527
Citation: CHEN Siyu, LIU Xianzhao, WANG Yixiang, et al. Individual tree detection in the high canopy density Larix principis-rupprechtii forest based on airborne LiDAR[J]. Journal of Zhejiang A&F University, 2022, 39(4): 800-806. DOI: 10.11833/j.issn.2095-0756.20210399
shu

    HTML

  • 华北落叶松Larix principis-rupprechtii是中国华北地区第一大树种，具有较高的经济和生态价值，探讨华北落叶松单木参数的提取、林分蓄积量的估算具有重要意义[1]。高精度的单木识别是获取单木和森林结构参数的重要前提，单木识别包括单木位置识别和树冠分割[2-4]。相比高分辨率卫星影像和光学影像，机载LiDAR技术不受区域信号影响，分辨率较高，是一种主动遥感技术，能获取林木的水平和垂直结构信息[5-6]。利用机载LiDAR研究单木识别的方法主要有2类：基于冠层高度模型(CHM)识别方法和基于归一化后的点云识别方法[3]。其中基于CHM的单木位置识别方法即在灰度图像中探测单木树顶，主要实现算法包括局部最大值法和多尺度分割法。局部最大值法假设树冠顶点高于树冠边缘，通过设置固定尺寸[7-10]或动态尺寸[11-12]的搜索窗口从CHM中识别树冠顶点特征，单木位置探测率可达85%。多尺度分割法以像元为单位，根据相似特征逐渐外扩分割，关键在于最优分割尺度的选取[13-17]。基于归一化后的点云单木位置识别方法以点云的结构信息为基础，主流实现算法包括区域生长法和均值漂移法。区域生长法通过选择合适的种子点，定义距离生长准则，判别不同属类[18-20]，选择的种子点和制定的距离生长规则不同，取得的单木识别精度也不同。均值漂移法是一种基于密度梯度上升的非参数方法，通过迭代运算求得同簇的数据点[21]，关键在于漂移均值向量的计算。单木识别时可利用林分树冠的距离和多维特征空间为向量来完成[22-23]，其在中低郁闭度林分中的单木识别精度较高，但在高郁闭度林分条件下，由于树冠重叠度增加，漂移均值向量计算会有偏差，识别结果不尽合理。为提高识别精度，结合树冠冠型特点，本研究提出基于机载LiDAR点云空间特征的高斯核函数改进的均值漂移单木位置识别方法(MSP)。为验证方法的有效性，对比了基于点云空间特征的区域生长图像分割方法(RGP)、基于CHM的局部最大值单木探测方法(LMC)和基于CHM的多尺度分割方法(MSC)等3种目前应用较为广泛的单木位置识别方法。最后，在单木位置识别结果上，进行单木树冠的提取。

    • 1.   研究区概况

    • 忻州市五寨县羊圈沟林场(38º44′~39º01′N，111º48′~112º03′E)位于山西省中北部，属亚寒带大陆性气候，季节变化明显，海拔为1 350~2 700 m，年均气温为5 ℃，年降水量为453.9 mm；华北落叶松-云杉 Picea asperata天然次生林与华北落叶松人工林带状混交。本研究在华北落叶松人工林带内设置8个20 m×20 m的正方形样地，植株密度为1 875~2 925 株·hm−2，郁闭度为0.8~1.0，是典型高郁闭度森林。

    2.   数据采集与研究方法

      2.1.   数据采集

    • 标记并编号8个样地内胸径大于5 cm 的华北落叶松，调查记录每木冠幅、树高和胸径；利用实时动态定位和全站型电子测距仪获取每木的位置坐标。样地林分基本情况如表1。

      样地密度/(株·hm−2)平均树高/m平均胸径/cm平均冠幅/m样地密度/(株·hm−2)平均树高/m平均胸径/cm平均冠幅/m
      12 07513.1113.223.1952 10011.7213.823.82
      22 92512.6612.442.8662 47512.0012.983.43
      32 50014.0415.352.9972 27513.4514.632.92
      41 87513.6013.413.9782 12513.5213.903.34

      Table 1.  Investigation results of basic characteristics of forest stand in sample plot

      2018年6月2日，使用GV 1500六旋翼无人机搭载RIGEL VUX-1激光扫描仪获取机载LiDAR点云数据；设置飞行高度为200 m，飞行速度为6 m·s−1，离散点云平均密度为70个·m−2，坐标系为WGS_1984_UTM_Zone_49N。

      在同一天，使用GV 120六旋翼无人机搭载传感器SONY Alpha7 RⅡ采集可见光遥感影像。通过Limapper V 2.0软件生成数字正射影像作为辅助数据，用于目视解译与可视化数据展示。

      • 2.2.   数据预处理

    • 首先，对原始点云数据进行2 m缓冲带裁剪，获取8个样地。然后，对样地点云数据进行以下处理：①采用距离阈值法移除噪声点。②采用渐进加密三角网滤波算法提取地面点云。为提高后期生成数字高程模型(DEM)的精度，设置迭代距离与迭代角度为1.4 m和8º，以便分离地面点云和非地面点云。③对去噪和分类后的点云数据进行克里金插值[23]。通过定义该点邻域中采样点的距离加权平均计算像元值。对地面点云和非地面点云分别插值，得到DEM和数字地表模型(DSM)，两者相减获得冠层高度模型(CHM) (图1)。图1C中，高程值较低的条带为华北落叶松人工林带，较高的为华北落叶松-云杉天然次生林。④基于LiDAR360 V 3.2软件，借助DEM对去噪后的点云进行高度信息归一化，避免地形起伏对点云高程的影响。

      Figure 1.  Study data

      • 2.3.   单木位置识别

      2.3.1.   基于点云空间特征的单木位置识别

    • 均值漂移算法是一种基于非参数概率密度梯度估计的迭代聚类过程[24-25]。单木点云分布具有与高斯核函数相似的几何形状特征，可用高斯核函数改进的均值偏移算法对点云数据进行聚类，进而获取单木位置。设置迭代运算的收敛阈值(ε)，基于点云空间特征的改进均值漂移单木位置识别方法，算法如下：(1) 在点云数据点中随机选择1个点作为初始中心点，找出离中心点距离在带宽h之内的所有点，记做集合M，认为这些点属于同一类簇；(2)选用点云三维空间特征，计算中心点到集合M中每个数据点的向量之和，得到偏移向量Mh(x)；(3)令中心点沿偏移向量方向移动，移动距离是偏移向量Mh(x)的模；(4)重复步骤(2)~(3)，直到偏移向量小于ε，记住此时的中心点；(5)重复步骤(1)~(4)，直到所有数据点都被归类；(6)提取单木位置，每一个类簇的中心点即为单木位置。

      • 2.3.2.   基于CHM的单木位置识别

    • 局部最大值法基于单木树冠顶部高程大于树冠边缘高程的假设，设置搜索窗口进行局部最大值探测。本研究将搜索窗口的起始尺寸设置为林分的平均冠幅大小，对窗口尺寸按增加或减少1个像素进行动态调整，搜索出树冠的局部最大值。基于CHM的图像分割法通过图像分割实现单木分割，根据区域中心确定单木位置。标记控制分水岭分割法是最常用的分割方法，但在本研究中分割精度不理想，因此选用面向对象的多尺度分割。利用eCognition Developer V 8.9软件的Estimation of Scale Parameter 2 (ESP2)工具包得到合适的分割尺度。本研究中，分割尺度为66，形状参数为0.8，紧致度参数为0.9，样地分割效果最好。

      • 2.4.   单木树冠提取

    • 在单木位置识别的基础上，划分出同一类点云的投影范围即为该株华北落叶松的树冠范围。考虑到冠层提取过程中，林分树冠间隙部分容易产生过分割现象，需要对其结果进行区域合并[26]，本研究将形状异质性作为类别之间的合并准则。选择eCognition软件对点云初始分类结果进行区域合并，分割尺度为58，形状参数和紧致度参数分别为0.8和0.8。

      • 2.5.   精度评价

      2.5.1.   单木位置识别精度评价

    • 根据卫星定位测量获取坐标的定位精度，考虑树顶与树根的位置偏差，将华北落叶松位置误差预计为1 m；以真值坐标点为圆心，创建半径为1 m的缓冲区。单木位置识别结果落入此缓冲区内，则认为正确。测算正确识别的单木总数(NC)、实地调查单木实际总数(NF)、识别出的单木总数(ND)、重复识别的单木总数(NR)等数据。计算精度评价指标，并使用单因素方差分析检验不同单木位置识别方法在各个指标方面的效果。①总体精度(AO)即正确识别的单木总数占实地调查单木实际总数的比值，公式为：AO=NC/NF×100%。②识别精度(AD)即正确识别的单木总数占识别出的单木总数的比值，公式为：AD=NC/ND×100%。③漏分误差(EO)即未识别到的单木总数占实地调查单木实际总数的比值，公式为：EO=(NF−NC)/NF×100%。④错分误差(EC)即被错误识别为单木的总数占识别出的单木总数的比值，公式为：EC=(ND−NC)/ND×100%。⑤重复性误差(ER)即对同一株单木重复识别的总数占识别出的单木总数的比值，公式为：ER=NR/ND×100%。

      • 2.5.2.   单木树冠提取精度评价

    • 采用单株冠幅提取精度(ω)来进行评价，公式为：ω=100%−∣(C1−C0)/C0∣×100%。其中：C0为实测冠幅，C1为单木对应的提取冠幅。

    3.   结果与分析

      3.1.   不同单木位置识别方法的识别效果

    • 利用4种方法进行单木位置识别，空间分布图(图2)显示：4种方法均存在漏分、错分和重复性识别的现象，结合表2可知：4种方法的识别精度(AD)由高到低依次为MSP (89.30%)、LMC (85.60%)、RGP (77.50%)、MSC (70.00%)；MSP法的单木位置识别效果要好于其他3种方法。4种方法的识别精度均小于总体精度(AO)，说明存在单木错误识别现象；相比之下，MSP法错分误差最低(EC为8.0%),，MSC法错分误差值最高(EC为30.1%)，说明依靠邻近相似特征识别单木的方法更容易产生将单木识别错误。相比RGP和MSC法，MSP法的漏分误差(EO)、错分误差(EC)和重复性误差(ER)均较低，说明高郁闭度针叶林中，按照邻域分属识别单木位置会受到树冠重叠度的影响，并不适用华北落叶松的单木位置识别。尽管LMC法和MSP法的总体精度都高于90%，但后者错分误差(Ec)较低，说明MSP法更能避免将非树对象错误识别为单木而增加识别总数的现象。

      Figure 2.  Spatial distribution of the result of individual tree position detection based on four methods

      类别方法AO/%AD/%EO/%EC/%ER/%
      点云MSP91.3±3.7 a89.3±2.5 a8.7±3.7 b8.0±2.5 c5.8±1.5 b
      RGP81.5±2.6 b77.5±2.2 b18.5±2.6 a16.7±5.5 b9.8±1.4 a
      CHMLMC91.0±2.0 a85.6±2.0 c9.0±2.0 b9.3±1.7 c4.3±1.7 b
      MSC80.8±2.1 b70.0±1.5 d19.2±2.1 a30.1±5.0 a5.0±1.4 b
      　　说明：同列不同字母表示差异显著(P＜0.05)

      Table 2.  Accuracy statistics of four methods on individual tree detection

      • 3.2.   改进均值漂移法的树冠提取效果

    • 采用均值漂移方法提取华北落叶松单株树冠(图3)。以单株华北落叶松实测冠幅为真值，对利用MSP法提取的单株冠幅进行精度分析，结合表3可知：MSP法总体平均提取精度为90.18%。单株提取精度小于80%的占总体结果的10.20%，单株提取精度大于80%的占总体结果的89.80%。单株树冠分割偏小的结果占总体的62.07%，要远远高于提取冠幅偏大的结果。

      Figure 3.  Crown extraction results of improved mean shift method (local)

      单株冠幅提取
      精度范围/%      		精度范围
      占比/%      		冠幅偏
      小/%      		冠幅偏
      大/%      		平均单株
      提取精度/%
      ＜60062.0737.9390.18
      60~701.05
      70~809.15
      80~9032.68
      90~10057.12

      Table 3.  Precision distribution of extracting individual canopy diameter by MSP

    4.   讨论与结论

      4.1.   讨论

    • 本研究提出的改进均值漂移法对高郁闭度华北落叶松林的单木提取总体精度达91.3%，这与张怡卓等[25]对中低密度人工林的研究结果(总体精度达89.0%)相似，远高于JING等[27]对加拿大高郁闭林分的总体精度(69.0%)。相比于其他3种提取方法，MSP法有2个优势。一是在林木树冠重叠的情况下，MSP法直接利用了点云空间特征，避免了生成CHM过程中产生的栅格误差；二是该方法利用统计迭代的方式确定聚类点位，根据访问频率分类，比直接比较邻域特征而进行分类稳健性更好。

      本研究发现：MSP法提取单木冠幅的平均提取精度为90.18%，比同等郁闭度林分的树冠提取精度要高出很多[28]，可满足森林调查需求。约10%的单株冠幅提取精度低于80%，这些点主要位于林分下层。由于上层遮挡，下层点云密度远低于上层，而点云聚类算法和点云密度息息相关，这也是本研究发现冠幅提取值偏小的原因。具体来说，由于上层冠层存在重叠，人工目视解译可以考虑到冠层重叠的影响并准确提取冠幅，而在三维点云图上，树冠界限分类时存在距离阈值类似的情况，影响提取结果；实测中，研究者往往通过东—西、南—北方向来测量冠幅，但是这只是一种理想状态下的测量，在现实中，华北落叶松林木树冠是不规则的，测量因此会存在误差；同时研究区域的坡度、地形等的限制，也会对人工测量造成影响。

      • 4.2.   结论

    • 本研究基于点云单一空间特征，引入高斯核函数简化并改进了均值漂移量的计算过程，同时所提方法不需要先验知识，算法的稳健性高，解决了高郁闭度的华北落叶松林受复杂冠层影响的问题，提高了在该林分内机载LiDAR单木识别的精度，为高郁闭度华北落叶松林单木识别提供了新的方法。

Reference (28)

Catalog

  • HTML

  • 1.   研究区概况
  • 2.   数据采集与研究方法

  • 2.1.   数据采集

  • 2.2.   数据预处理

  • 2.3.   单木位置识别

  • 2.3.1.   基于点云空间特征的单木位置识别
  • 2.3.2.   基于CHM的单木位置识别

  • 2.4.   单木树冠提取

  • 2.5.   精度评价

  • 2.5.1.   单木位置识别精度评价
  • 2.5.2.   单木树冠提取精度评价
  • 3.   结果与分析

  • 3.1.   不同单木位置识别方法的识别效果

  • 3.2.   改进均值漂移法的树冠提取效果

  • 4.   讨论与结论

  • 4.1.   讨论

  • 4.2.   结论

浙ICP备11046845号

Copyright by 2019 Editorial Department of Journal of Zhejiang A&F University

Address: No. 666, Wu Su street, Ling'an District, Hangzhou, ZhejiangPost Code: 311300Tel: 0571-63732749

E-Mail: :zlxb@zafu.edu.cn

Supported by: Beijing Renhe Information Technology Co. Ltd  support: info@rhhz.net

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return