Predicting DBH of a single <i>Ginkgo biloba</i> tree based on UAV images

JIA Penggang; XIA Kai; DONG Chen; FENG Hailin; YANG Yinhui

doi:10.11833/j.issn.2095-0756.2019.04.016

Volume 36 Issue 4

Jul. 2019

Turn off MathJax

Article Contents

Article Navigation > Journal of Zhejiang A&F University > 2019 > 36(4): 757-763

JIA Penggang, XIA Kai, DONG Chen, FENG Hailin, YANG Yinhui. Predicting DBH of a single Ginkgo biloba tree based on UAV images[J]. Journal of Zhejiang A&F University, 2019, 36(4): 757-763. doi: 10.11833/j.issn.2095-0756.2019.04.016

Citation:

JIA Penggang, XIA Kai, DONG Chen, FENG Hailin, YANG Yinhui. Predicting DBH of a single Ginkgo biloba tree based on UAV images[J]. Journal of Zhejiang A&F University, 2019, 36(4): 757-763. doi: 10.11833/j.issn.2095-0756.2019.04.016

Predicting DBH of a single Ginkgo biloba tree based on UAV images

doi: 10.11833/j.issn.2095-0756.2019.04.016

JIA Penggang^{1,2
,},
XIA Kai^{1,2
,
,},
DONG Chen^1,2,
FENG Hailin^1,2,
YANG Yinhui^1,2

1.
School of Information Engineering, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China
2.
Zhejiang Provincial Key Laboratory of Forestry Intelligent Monitoring and Information Technology, Zhejiang A&F University, Hangzhou 311300, Zhejiang, China

Received Date: 2018-08-29
Rev Recd Date: 2018-12-03
Publish Date: 2019-08-20

Abstract

To efficiently calculate and predict forest stock and biomass, diameter at breast height (D_BH), a basic factor of a tree, was used in a regression model. In this study, Ginkgo biloba was used as the research object. Image data was obtained with an unmanned aerial vehicle (UAV), and using the method of structure from motion (SFM), a digital surface model and an orthophoto map were generated. Next, the canopy area (A_C), crown width (W_C) and tree height (H) of G. biloba were extracted. Then, one-way regression models (A_C-D_BH, WC-D_BH, H-D_BH), binary regression models (A_C&W_C-D_BH, A_C&H-D_BH, W_C&H-D_BH), and a ternary regression model (A_C&W_C&H-D_BH) were established. Results of 52 groups of fitted samples showed that the A_C&W_C&H-D_BH model had the highest coefficient of determination (R² = 0.825 0) and the lowest root mean square error (E_RMS = 0.959 1). Results of 19 groups of test samples showed that the D_BH error rate for the A_C&W_C&H-D_BH model was 4.20%, which was less than the allowable error value (5%) for the A-type forest resource D_BH factor. Thus, a high precision D_BH value could be calculated using the three parameters of canopy area, crown width, and tree height, thereby providing a new idea for automated forest resource surveying and monitoring.
- forest mensuration,
- unmanned aerial vehicle (UAV),
- diameter at breast height (DBH),
- crown area,
- crown width,
- height,
- inversion model

Relative Article

[1]	LAN Zhenzhen, WANG Xinpu, SHI Xingwen, DONG Chuan. Key factors affecting the occurrence of Acantholyda posticalis larvae in Luoshan Mountains, Ningxia . Journal of Zhejiang A&F University, 2024, 41(3): 1-9. doi: 10.11833/j.issn.2095-0756.20230581
[2]	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(4): 1-10. doi: 10.11833/j.issn.2095-0756.20240107
[3]	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
[4]	WU Danzi, WANG Chengde, LI Liang, LIU Min. Crown profile and volume compatibility model of Cunninghamia lanceolata in Fujian Province . Journal of Zhejiang A&F University, 2020, 37(1): 114-121. doi: 10.11833/j.issn.2095-0756.2020.01.015
[5]	MAO Guangxian, TAN Wei, CHAI Zongzheng, ZHAO Yang, YANG Shenjun. Diameter-height model for Pinus massoniana plantations based on BP neural network . Journal of Zhejiang A&F University, 2020, 37(4): 752-760. doi: 10.11833/j.issn.2095-0756.20190486
[6]	XIE Qiaoya, YU Kunyong, DENG Yangbo, LIU Jian, FAN Huadong, LIN Tongzhou. Height measurement of Cunninghamia lanceolata plantations based on UAV remote sensing . Journal of Zhejiang A&F University, 2019, 36(2): 335-342. doi: 10.11833/j.issn.2095-0756.2019.02.015
[7]	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
[8]	LUO Hengchun, ZHANG Chao, WEI Anchao, ZHANG Yi, HUANG Tian, YU Zhexiu. Average DBH growth model of a stand with environmental parameters for Pinus yunnanensis in central Yunnan, China . Journal of Zhejiang A&F University, 2018, 35(6): 1079-1087. doi: 10.11833/j.issn.2095-0756.2018.06.011
[9]	LEI Zeyong, HAN Yangang, ZHAO Guojun, ZHOU Yanping, ZHANG Yansong, YU Deliang. Growth of Mongolian pine in Zhanggutai of Liaoning Province . Journal of Zhejiang A&F University, 2018, 35(2): 324-330. doi: 10.11833/j.issn.2095-0756.2018.02.017
[10]	WANG Mingchu, SUN Yujun. Based on mixed-effects model and empirical best linear unbiased predictor predicting growth profile of height for Chinese fir . Journal of Zhejiang A&F University, 2017, 34(5): 782-790. doi: 10.11833/j.issn.2095-0756.2017.05.003
[11]	CHEN Jinxing, YUE Depeng, FENG Zhongke, DING Jiawei, YAO Bingquan, YE Tianxiong. A handheld dendrometer for automatic tree diameter measurement . Journal of Zhejiang A&F University, 2016, 33(4): 589-598. doi: 10.11833/j.issn.2095-0756.2016.04.006
[12]	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
[13]	CHEN Ting, SHI Yongjun, ZHOU Guomo, ZHENG Zerui, LI Chong. Interannual variation characteristics for stand structures in the early stages of new moso bamboo carbon sink stands . Journal of Zhejiang A&F University, 2015, 32(2): 181-187. doi: 10.11833/j.issn.2095-0756.2015.02.003
[14]	SHANG Zhenzhen, ZHOU Guomo, DU Huaqiang. Relationship between above-ground biomass and DBH for Phyllostachys edulis stands based on fractal theory . Journal of Zhejiang A&F University, 2013, 30(3): 319-324. doi: 10.11833/j.issn.2095-0756.2013.03.002
[15]	WANG Xiao-ning, XU Tian-shu, LI Yi. Estimating forest volume in hilly regions with the ALOS PALSAR model’s dual polarization data . Journal of Zhejiang A&F University, 2012, 29(5): 667-670. doi: 10.11833/j.issn.2095-0756.2012.05.005
[16]	MA Rui-sheng, YANG Bin, ZHANG Li-hui, LIU Zhi-ping. Design and implementation of a forest fire monitoring system using a miniature unmanned aerial vehicle . Journal of Zhejiang A&F University, 2012, 29(5): 783-789. doi: 10.11833/j.issn.2095-0756.2012.05.023
[17]	WEI Xiao-hui, SUN Yu-jun, MA Wei. A height growth model for Cunninghamia lanceolata based on Richards’ equation . Journal of Zhejiang A&F University, 2012, 29(5): 661-666. doi: 10.11833/j.issn.2095-0756.2012.05.004
[18]	YANG Qian-yu, XIE Jin-zhong, ZHANG Wei, LIN Zhen-qing. Biomass models for Bambusa textilis var. tasca . Journal of Zhejiang A&F University, 2011, 28(3): 519-526. doi: 10.11833/j.issn.2095-0756.2011.03.027
[19]	MA Feng-feng, JIA Li-ming. Outside and inside bark diameter at breast height for Platycladus orientalis and Pinus tabulaeformis in the Beijing area . Journal of Zhejiang A&F University, 2009, 26(1): 13-16.
[20]	Xie Zhegen, Liu Anxing, Chen Xuetang, Chen Antong, Xu Zufu.. Study on Relative Tree Height Models. . Journal of Zhejiang A&F University, 1998, 15(1): 69-75.

References

[1]	XU Weiheng, FENG Zhongke, SU Zhifang, et al. An automatic extraction algorithm for individual tree crown projection area and volume based on 3D point cloud data[J]. Spectrosc Spectral Anal, 2014, 34(2):465-471.
[2]	PANAGIOTIDIS D, ABDOLLAHNEJAD A, SUROVY P. Determining tree height and crown diameter from high-resolution UAV imagery[J]. Int J Remote Sensing, 2017, 38(8/10):2392-2410.
[3]	BRANDTBERG T, WARNER T A, LANDENBERGER R E, et al. Detection and analysis of individual leaf-off tree crowns in small footprint, high sampling density LiDAR data from the eastern deciduous forest in North America[J]. Remote Sensing Environ, 2003, 85(3):290-303.
[4]	VASTARANTA M, LATORRE E G, LUOMA V, et al. Evaluation of a Smartphone APP for forest sample plot measurements[J]. Forests, 2015, 6(4):1179-1194.
[5]	BU Guochao, WANG Pei. Adaptive estimation method for diameter at breast height based on terrestrial laser scanning[J]. Laser Optoelectronics Prog, 2016, 53(8):278-286.
[6]	HUANG Shaoming, TITUS S J, WIENS D P. Comparison of nonlinear height diameter functions for major Alberta tree species[J]. Can J For Res, 1992, 22(9):1297-1304.
[7]	SOARES P, TOMÉ M. Height-diameter equation for first rotation eucalypt plantations in Portugal[J]. For Ecol Manage, 2002, 166(1):99-109.
[8]	TRINCADO G, VANDERSCHAAF C L, BURKHART H E. Regional mixedeffects height-diameter models for loblolly pine (Pinus taeda L.) plantations[J]. Eur J For Res, 2007, 126(2):253-262.
[9]	DE'ATH G. Multivariate regression trees:a new technique for modeling species-environment relationships[J]. Ecology, 83(4):1105-1117.
[10]	WANG Dongzhi, ZHANG Dongyan, ZHANG Zhidong, et al. Height-diameter relationship for conifer mixed forest based on nonlinear mixed-effects model[J]. Sci Silv Sin, 2016, 52(1):30-36.
[11]	DONG Chen, WU Baoguo, ZHANG Han. Parametric prediction models of DBH and height for Cunninghamia lanceolata plantation based on crown width[J]. J Beijing For Univ, 2016, 38(3):55-63.
[12]	HE Youyun, ZHANG Yubo, LI Junqing, et al. Estimation of steam biomass of individual Abies faxoniana through unmanned aerial vehicle remote sensing[J]. J Beijing For Univ, 2016, 38(5):42-49.
[13]	PENA J M, TORRES-SÁNCHEZ J, SERRANO-PÉREZ A, et al. Quantifying efficacy and limits of unmanned aerial vehicle (UAV) technology for weed seedling detection as affected by sensor resolution[J]. Sensors, 2015, 15(3):5609-5626.
[14]	BI Kai, LI Yingcheng, DING Xiaobo, et al. Aerial photogrammetric technology of light small UAV:status and trend of development[J]. Bull Surv Mapp, 2015(3):27-31, 48.
[15]	HERNANDEZ J G, FERREEIRO E G, SARMENTO A, et al. Using high resolution UAV imagery to estimate tree variables in Pinus pinea plantation in Portugal[J]. For Syst, 2016, 25(2):eSC09. doi:10.5424/fs/2016252-08895.
[16]	DEMETRIOS G, JEREMYS F, VICENTES M. Challenges to estimating tree height via LiDAR in closed-canopy forests:a parable from western Oregon[J]. For Sci, 2010, 56(2):139-155.
[17]	LIU Wenping, ZHONG Tingyu, SONG Yining. Prediction of trees diameter at breast height based on unmanned aerial vehicle image analysis[J]. Trans Chin Soc Agric Eng, 2017, 33(21):99-104.

Proportional views

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

Figures(2) / Tables(4)

Get Citation

PDF

XML

Article views(2986) PDF downloads(58) Cited by()

Proportional views

HTML

胸径（D_BH），树高（H），树冠面积（即树冠投影面积^[1]A_C）及冠幅（W_C）是进行森林资源清查的重要参数，是计算森林生物量、蓄积量的基础。传统森林资源调查主要通过人工检尺获得胸径值，这种方法具有成本高、效率低、数据滞后等弊端。遥感技术和激光雷达技术的发展与完善为森林信息的提取评估提供了新的方法^[2]。研究人员可以利用光检测和激光测量来估算林分层次上个体树的信息^[3]，新设备如激光测距仪和智能手机^[4]的使用提高了数据收集的效率和精度。步国超等^[5]为了自动化地、准确地从单站地面激光雷达数据中提取一定范围内的树木胸径，提出一种基于点云切片的圆形-椭圆自适应胸径估计方法。该方法可以快速有效地进行自适应胸径估计。国外学者较早通过建立线性或非线性预估模型的方法来反演单木参数^[6-8]，如DE’ATH^[9]通过建立树高—胸径回归方程预测不同样地树高。近年来，随着森林价值的不断挖掘，建立单木树冠面积、树高等因子与胸径的反演模型成为国内学者的研究重点。王冬至等^[10]建立了树高—胸径生长关系非线性混合效应模型，为研究混交林多树种生长规律提供参考依据。董晨等^[11]以杉木Cunninghamia lanceolata为研究对象，使用参数预估法构造胸径和树高的参数化预估模型。何游云等^[12]采用人机交互的方式提取单木树冠面积数据，并结合野外实测的胸径，建立树冠面积—胸径估算模型。结果显示：两者存在较好的非线性相关关系。无人机（UAV）是利用无线电遥控设备和自备的程序控制装置操纵的不载人飞行器，具有体积小、速度快、使用方便等优势，便于进行跟踪、定位、遥测和数字传输等工作，成为传统航空摄影测量手段的有效补充^[13-14]。由于无人机可以携带高分辨率相机，能够满足森林资源调查中高空间分辨率的要求，学者开始关注利用无人机获取具体林木信息以及森林信息量化等方面的工作。无人机遥感的影像分辨率大大超越了卫星遥感，在获取单木参数上具有天然优势，相关研究不断涌现。HERNANDEZ等^[15]基于无人机影像采用混合像元法估测出了单木树高、冠幅等参数；DEMETRIOS等^[16]使用无人机搭载的激光雷达对树冠覆盖度高的雨林进行树高测量。刘文萍等^[17]通过拟合无人机正射图像中的单木树冠面积与胸径的关系预测树木胸径值。前述研究表明：以树高，冠幅等参数反演胸径的方法基本成立，但传统手工量测提取单木参数的方法工作量大，难以应用于实践，而目前基于无人机影像的胸径预测研究较少，也不够全面。基于此，本研究以银杏Ginkgo biloba为研究对象，采用无人机获取样地影像数据，以树冠面积，冠幅和树高为自变量分别建立一元回归模型、二元回归模型和三元回归模型，目的是证明无人机遥感影像数据可用于胸径预估，并探索最佳的反演模型，推进森林资源调查研究的自动化。

1. 研究方法

1.1. 研究区概况

研究区位于浙江省杭州市浙江农林大学校园内，30.25°~30.26°N，119.72°~119.73°E，亚热带季风气候，四季分明，夏季温润多雨，冬季干燥少雨；校园内有300余种乔木，以亚热带常绿阔叶林为主。从谷歌地球（Google Earth）上截取研究区范围，共选取4块银杏样地作为试验区，每块样地为15 m × 150 m。其中样地1，样地2和样地3用来做模型拟合，共52组有效数据；样地4用来做模型反演精度检验，共19组有效数据。

1.2. 无人机系统

本研究所使用的无人机为大疆悟INSPIRE 2系列，主要由4部分组成：飞行器、遥控器、云台相机、电源。飞行器的型号为T650，质量为3 290 g（含2块电池），适用工作环境温度：-10~40 ℃，双电池系统可实现最长飞行时间约25 min。无人机搭载的云台相机型号为大疆ZenmuseX5S，镜头规格为DJI MFT 15 mm，有效像素为2 080万（表 1）。在飞行中搭配DJI GO PRO软件使用，主要是用于样地飞行路线规划。

	指标	参数
飞行器	类型	四轴飞行器
	飞行载质量	4 kg
	飞行速度	最大水平飞行速度：108 km·h^-1
	飞行高度	最大起飞海拔高度：2 500 m
遥控器	工作频率	2.400~2.483 GHz；5.725~5.850 GHz
	控制距离	2.4 GHz：7.0 km(FCC)；3.5 km(CE)；4.0 km(SRRC) 5.8 GHz：7.0 km(FCC)；2.0 km(CE)；5.0 km(SRRC)
云台相机	型号	大疆ZenmuseX5S
	外形尺寸	140 mm × 98 mm × 132 mm
	产品质量	461 g
	有效像素	2 080万
	图像分辨率	4:3 5 280 × 3 956
		16:9 5 280 × 2 970
	可控转动范围	俯仰：30°一90°

Table 1. Dajiang Wu INSPIRE 2 UAV parameters

1.3. 技术路线

本研究基于无人机遥感影像研究单木胸径反演模型。具体流程为：首先利用无人机进行样地影像的拍摄，同时实测样地内单木的胸径；利用运动恢复结构（structure from motion，SFM）方法进行三维重建并生成数字表面模型（digital surface model，DSM）和正射影像图（digital orthophoto map，DOM）；利用ENVI Lidar读取数字表面模型点云数据并提取树高；通过Arc GIS 10.1获得树冠的面积值和冠幅值，进而根据胸径、树高、冠幅和冠幅面积4个林木因子，分别建立一元回归模型（A_C-D_BH, W_C-D_BH, H-D_BH），二元回归模型（A_C&W_C-D_BH, A_C&H-D_BH, W_C&H-D_BH）和三元回归模型（A_C&W_C&H-D_BH），最后通过误差检验获得最佳反演模型。

4. 结论

本研究以银杏为研究对象，通过无人机获取影像数据并提取树冠面积、冠幅和树高3个参数，与胸径建立了多个反演模型。其中以冠幅、树冠面积和树高为自变量反演胸径的三元一次模型的决定系数为0.825 0，均方根误差为0.959 1，19组验证数据的平均误差值为0.76 cm，误差率为4.2%，是最优的模型，满足A类森林资源调查胸径因子误差小于5%的规定，可以应用到森林资源调查的实践中。另一方面，二元二次模型W_C&H-D_BH的拟合精度较高，反演误差率较低，在实际应用中，可根据客观条件选择合适的模型。本研究证明了无人机遥感影像数据可以用于胸径预估，并找到了单木银杏胸径的最佳反演模型。下一步的工作是通过编程实现树冠面积、冠幅和树高值的提取，进而推动森林资源调查的自动化。

由于本研究是在特定立地条件下进行的，而复杂地形（例如山区）或者具体密集林分地区可能会在一定程度上影响实验结果，如何提高复杂地形或者具体密集林分地区单木信息的提取精度是需要进一步解决的问题。同时也要指出，本研究仅针对较小区域内的一个树种进行多元胸径反演模型研究，不同区域或者不同树种可能会影响研究结果。

Reference (17)

[1]	徐伟恒, 冯仲科, 苏志芳. 1种基于三维激光点云数据的单木树冠投影面积和树冠体积自动提取算法[J]. 光谱学与光谱分析, 2014, 34(2): 465-471. doi: 10.3964/j.issn.1000-0593(2014)02-0465-07	XU Weiheng, FENG Zhongke, SU Zhifang. An automatic extraction algorithm for individual tree crown projection area and volume based on 3D point cloud data[J]. Spectrosc Spectral Anal, 2014, 34(2): 465-471. doi: 10.3964/j.issn.1000-0593(2014)02-0465-07
[2]	PANAGIOTIDIS D, ABDOLLAHNEJAD A, SUROVY P. Determining tree height and crown diameter from high-resolution UAV imagery[J]. Int J Remote Sensing, 2017, 38(8/10): 2392-2410.
[3]	BRANDTBERG T, WARNER T A, LANDENBERGER R E. Detection and analysis of individual leaf-off tree crowns in small footprint, high sampling density LiDAR data from the eastern deciduous forest in North America[J]. Remote Sensing Environ, 2003, 85(3): 290-303. doi: 10.1016/S0034-4257(03)00008-7
[4]	VASTARANTA M, LATORRE E G, LUOMA V. Evaluation of a Smartphone APP for forest sample plot measurements[J]. Forests, 2015, 6(4): 1179-1194.
[5]	步国超, 汪沛. 基于单站地面激光雷达数据的自适应胸径估计方法[J]. 激光与光电子学进展, 2016, 53(8): 278-286.	BU Guochao, WANG Pei. Adaptive estimation method for diameter at breast height based on terrestrial laser scanning[J]. Laser Optoelectronics Prog, 2016, 53(8): 278-286.
[6]	HUANG Shaoming, TITUS S J, WIENS D P. Comparison of nonlinear height diameter functions for major Alberta tree species[J]. Can J For Res, 1992, 22(9): 1297-1304. doi: 10.1139/x92-172
[7]	SOARES P, TOMÉ M. Height-diameter equation for first rotation eucalypt plantations in Portugal[J]. For Ecol Manage, 2002, 166(1): 99-109.
[8]	TRINCADO G, VANDERSCHAAF C L, BURKHART H E. Regional mixedeffects height-diameter models for loblolly pine (Pinus taeda L.) plantations[J]. Eur J For Res, 2007, 126(2): 253-262. doi: 10.1007/s10342-006-0141-7
[9]	DE'ATH G. Multivariate regression trees:a new technique for modeling species-environment relationships[J]. Ecology, , 83(4): 1105-1117.
[10]	王冬至, 张冬燕, 张志东. 基于非线性混合模型的针阔叶混交林树高与胸径关系[J]. 林业科学, 2016, 52(1): 30-36. doi: 10.3969/j.issn.1006-1126.2016.01.006	WANG Dongzhi, ZHANG Dongyan, ZHANG Zhidong. Height-diameter relationship for conifer mixed forest based on nonlinear mixed-effects model[J]. Sci Silv Sin, 2016, 52(1): 30-36. doi: 10.3969/j.issn.1006-1126.2016.01.006
[11]	董晨, 吴保国, 张翰. 基于冠幅的杉木人工林胸径和树高参数化预估模型[J]. 北京林业大学学报, 2016, 38(3): 55-63.	DONG Chen, WU Baoguo, ZHANG Han. Parametric prediction models of DBH and height for Cunninghamia lanceolata plantation based on crown width[J]. J Beijing For Univ, 2016, 38(3): 55-63.
[12]	何游云, 张玉波, 李俊清. 利用无人机遥感测定岷江冷杉单木树干生物量[J]. 北京林业大学学报, 2016, 38(5): 42-49.	HE Youyun, ZHANG Yubo, LI Junqing. Estimation of steam biomass of individual Abies faxoniana through unmanned aerial vehicle remote sensing[J]. J Beijing For Univ, 2016, 38(5): 42-49.
[13]	PENA J M, TORRES-SÁNCHEZ J, SERRANO-PÉREZ A. Quantifying efficacy and limits of unmanned aerial vehicle (UAV) technology for weed seedling detection as affected by sensor resolution[J]. Sensors, 2015, 15(3): 5609-5626. doi: 10.3390/s150305609
[14]	毕凯, 李英成, 丁晓波. 轻小型无人机航摄技术现状及发展趋势[J]. 测绘通报, 2015, (3): 27-31, 48.	BI Kai, LI Yingcheng, DING Xiaobo. Aerial photogrammetric technology of light small UAV:status and trend of development[J]. Bull Surv Mapp, 2015, (3): 27-31, 48.
[15]	HERNANDEZ J G, FERREEIRO E G, SARMENTO A. Using high resolution UAV imagery to estimate tree variables in Pinus pinea plantation in Portugal[J]. For Syst, 2016, 25(2): eSC09-. doi: 10.5424/fs/2016252-08895
[16]	DEMETRIOS G, JEREMYS F, VICENTES M. Challenges to estimating tree height via LiDAR in closed-canopy forests:a parable from western Oregon[J]. For Sci, 2010, 56(2): 139-155.
[17]	刘文萍, 仲亭玉, 宋以宁. 基于无人机图像分析的树木胸径预测[J]. 农业工程学报, 2017, 33(21): 99-104. doi: 10.11975/j.issn.1002-6819.2017.21.012	LIU Wenping, ZHONG Tingyu, SONG Yining. Prediction of trees diameter at breast height based on unmanned aerial vehicle image analysis[J]. Trans Chin Soc Agric Eng, 2017, 33(21): 99-104. doi: 10.11975/j.issn.1002-6819.2017.21.012

项目	树冠面积（提取)/m²	冠幅（提取)/m	树高（提取)/m	树高（实测)/m	胸径（实测)/cm
最小值	4.08	3.02	4.8	5.2	13.8
最大值	21.78	6.25	13.9	12.3	23.1
平均值	12.97	4.28	9.5	8.7	18.5
标准差	4.84	0.74	1.81	1.39	2.08

	模型	类型	表达式	决定系数(R²)	均方根误差(E_RMS)
	A_C-D_BH	指数函数	D_BH=14.11e(0.020 89A_C)	0.729 9	1.168 0
		幂函数	D_BH=9.423A_C^0.2704	0.758 7	1.100 0
		一次多项式	D_BH=0.401 7A_C+13.37	0.743 2	1.140 0
		二次多项式	D_BH=-0.013 42A_C²+0.769 6Ac+11.15	0.761 8	1.100 0
	H-D_BH	指数函数	D_BH=12.56e(0.042 03H)	0.546 1	1.514 0
		幂函数	D_BH=7.615H^{0.403 7}	0.569 2	1.480 0
一元模型		一次多项式	D_BH=0.817 3H+11.02	0.560 4	1.490 0
		二次多项式	D_BH=-0.043 95H²+1.66H+7.16	0.575 3	1.480 0
	W_C-D_BH	指数函数	D_BH=11.78e^{(0.106 2W_C)}	0.534 1	1.534 0
		幂函数	D_BH=8.98W_C^{0.503 8}	0.579 3	1.458 0
		一次多项式	D_BH=2.141W_C+9.448	0.557 8	1.495 0
		二次多项式	D_BH=-0.928 1W_C²+10.68W_C-9.583	0.648 9	1.354 0
	A_C&W_C-D_BH	一次多项式	D_BH=0.613A_C-1.389W_C+16.58	0.772 3	1.084 0
		二次多项式	D_BH=-0.004 122A_C²+0.818 8A_C-0.631 1W_C-0.031 71A_CW_C+13.31	0.780 2	1.087 0
			D_BH=0.046 69W_C²-0.566 4W_C+0.841 3A_C-0.063 62A_CW_C+12.96	0.780 1	1.087 0
	W_C&H-D_BH	一次多项式	D_BH=0.447 5H+2.114W_C+5.624	0.807 6	1.069 0
二元模型		二次多项式	D_BH=-0.025 26H²+1.117H+2.519W_C-0.048 53HW_C+1.912	0.813 4	1.074 0
			D_BH=0.098 34W_C²+2.621W_C+1.062H-0.149W_CH+1.929	0.812 3	1.077 0
	A_C&H-D_BH	一次多项式	D_BH=0.310 7A_C+0.3H+11.75	0.780 6	1.064 0
		二次多项式	D_BH=-0.005 197A_C²+0.619 1A_C+0.517 8H-0.017 21A_CH+8.909	0.795 1	1.050 0
			D_BH=0.041 04H²+0.132H+0.762 5A_C-0.047 16HA_C+9.747	0.796 7	1.045 0
三元模型	A_C&W_C&H-D_BH	一次多项式	D_BH=0.55A_C-1.695W_C+0.374H+15.226	0.825 0	0.959 1
说明：A_C为树冠面积（m²），W_C为冠幅（m），H为树高（m）

	模型	表达式	平均误差率/I
	A_C-D_BH	D_BH=-0.0134 2A_C²+0.769 6A_C+11.15	5.34
一元模型	H-D_BH	D_BH=-0.043 95H²+1.66H+7.16	7.40
	W_C-D_BH	D_BH=-0.928 1W_C²+10.68W_C-9.583	6.74
	A_C&W_C-D_BH	D_BH=-0.004 122A_C²+0.818 8A_C-0.631 1W_C-0.0317 1A_CW_C+13.31	4.90
二元模型	W_C&H-D_BH	D_BH=-0.025 26H²+1.117H+2.519W_C-0.048 53HW_C+1.912	4.95
	A_C&H-D_BH	D_BH=0.041 04H²+0.132H+0.762 5A_C-0.047 16HA_C+9.747	4.27
三元模型	A_C&W_C&H-D_BH	D_BH=0-55A_C-1.695W_C+0.374H+15.226	4.20

Predicting DBH of a single Ginkgo biloba tree based on UAV images

doi: 10.11833/j.issn.2095-0756.2019.04.016