Please wait a minute...
img

官方微信

遥感技术与应用  2019, Vol. 34 Issue (5): 939-949    DOI: 10.11873/j.issn.1004-0323.2019.5.0939
林业遥感专栏     
集成U-Net方法的无人机影像胡杨树冠提取和计数
李越帅1,2(),郑宏伟1,2(),罗格平1,2,杨辽1,王伟胜1,桂东伟1
1. 中国科学院新疆生态与地理研究所 荒漠与绿洲国家重点实验室,新疆 乌鲁木齐 830011
2. 中国科学院大学,北京 100049
Extraction and Counting of Populus Euphratica Crown Using UAV Images Integrated with U-Net Method
Yueshuai Li1,2(),Hongwei Zheng1,2(),Geping Luo1,2,Liao Yang1,Weisheng Wang1,Dongwei Gui1
1. State Key Laboratory of Desert and Oasis Ecology, Xinjiang Istitute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China
2. University of Chinese Academy of Sciences, Beijing 100049, China
 全文: PDF(5095 KB)   HTML
摘要:

塔里木河流域的胡杨林是该荒漠区域典型的森林资源,胡杨树冠大小和株数信息对塔里木河流域森林资源监测、生态保护和恢复具有重要意义。由于该流域乔灌草植物群落分布的复杂性,传统方法很难实现胡杨树冠的精准分割和大范围的株数提取。以塔里木河中游胡杨林为研究区,选取几块典型胡杨林区域,提出集成深度学习和分水岭分割的处理方法,对密集胡杨树冠的精准分割和单株胡杨的提取进行了深入探讨。首先,将无人机影像(空间分辨率0.16 m)无缝拼接生成正射影像;采用U-Net卷积神经网络对胡杨树冠覆盖区域进行精准分割;在U-Net模型分割的基础上使用标记分水岭方法对密集胡杨树冠进行自动再分割和单株计数,计算出所选研究区的胡杨株数并精准定位。结果表明U-Net卷积神经网络对胡杨的所有树冠区域提取的平均精度可达94.1%,在胡杨树冠覆盖区域识别分割的基础上,用标记分水岭分割方法对胡杨单木计算总体精度为93.3%。研究认为,结合深度学习和标记分水岭方法为自动化大范围森林资源监测提供了新思路和借鉴经验。

关键词: 无人机影像胡杨深度学习分水岭树冠株数    
Abstract:

The Populus euphratica forest in the Tarim River Basin is a typical forest resource in the desert area. The canopy size and plant number information of Populus euphratica is of great significance for forest resource monitoring, ecological protection and restoration in the Tarim River Basin. Due to the complexity of the distribution of arbor, shrub and grass communities in the area, it is difficult to achieve accurate segmentation of canopy in dense Populus euphratica and large-scale plant number extraction. Taking the Populus euphratica forest in the middle of Tarim River as the research area, several typical Populus euphratica forest areas were selected, and the integrated processing methods of fusion deep learning and watershed segmentation were proposed. The precise segmentation of dense Populus euphratica and the extraction of Populus euphratica were carefully discussed in depth. First, the drone images (spatial resolution 0.16 m) are seamlessly stitched together to generate an orthophoto. Then U-Net convolutional neural network was used to accurately segment the canopy cover area of ??Populus euphratica. Furthermore, the marker segmentation method was used to automatically re-segment and count the intensive Populus canopy, and the number of Populus euphratica in the selected study area was calculated and accurately positioned. The results show that the average accuracy of the extraction of all canopy regions of Populus euphratica by integrated U-Net convolutional neural network is up to 94.1%. The overall accuracy of the calculation of Populus euphratica by the marker watershed segmentation method is 93.3%. The study suggests that the combination of deep learning and marker watershed methods can provide new ideas and lessons for the automation of large-scale forest resource monitoring.

Key words: UAV image    Populus euphratica    Deep learning    Watershed    Tree crown    Tree counting
收稿日期: 2018-12-06 出版日期: 2019-12-05
ZTFLH:  TP79  
基金资助: 国家重点研发计划“一带”核心区域生态环境安全监测与应急响应示范(2017YFB0504204);中国科学院特色研究所主要服务项目(TSS?2015?014?FW?1?3);国家自然基金面上项目(41877012)
通讯作者: 郑宏伟     E-mail: liyueshuai16@mails.ucas.ac.cn;hzheng@ms.xjb.ac.cn
作者简介: 李越帅(1992-),男,河南滑县人,硕士研究生,主要从事遥感图像信息提取研究。E?mail:liyueshuai16@mails.ucas.ac.cn
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
李越帅
郑宏伟
罗格平
杨辽
王伟胜
桂东伟

引用本文:

李越帅,郑宏伟,罗格平,杨辽,王伟胜,桂东伟. 集成U-Net方法的无人机影像胡杨树冠提取和计数[J]. 遥感技术与应用, 2019, 34(5): 939-949.

Yueshuai Li,Hongwei Zheng,Geping Luo,Liao Yang,Weisheng Wang,Dongwei Gui. Extraction and Counting of Populus Euphratica Crown Using UAV Images Integrated with U-Net Method. Remote Sensing Technology and Application, 2019, 34(5): 939-949.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2019.5.0939        http://www.rsta.ac.cn/CN/Y2019/V34/I5/939

图1  研究区及处理流程示意图
图2  树冠分割提取和株数计算流程图
图3  U-Net结构框架(蓝色条框代表多通道特征;灰色条框代表复制多通道特征;条框顶部数字代表通道数目;箭头代表不同的操作)
图4  树冠中心点检测出现的几种情况
图5  胡杨树冠分割结果
图6  不同方法的树冠提取结果比较

测试

区域

总体精度/%召回率/%IoU/%Kappa
ObiectSVMU-NetOOSVMU-NetObiectSVMU-NetObiectSVMU-Net
区域185.490.094.877.488.886.455.167.379.50.610.740.85
区域288.091.495.670.677.292.160.570.284.60.680.770.89
区域388.289.793.468.893.188.364.373.680.40.700.770.84
区域486.791.492.667.481.192.357.571.576.90.640.780.82
均值87.190.694.171.085.089.859.370.680.40.660.760.85
表1  不同方法树冠分割精度评价
Fig.7  The computing workflow of our proposed watershed segmentation method图7实验步骤(红色点表示密集区单木位置,绿色点表示稀疏区单木位置)
样本编号验证数/株尺寸4 /株尺寸5/株尺寸6/株尺寸7/株尺寸8 /株尺寸9/株
1116135114100908477
264988170696460
356827056494743
48113110583747063
570998171686256
6641159574706357
7791119185766965
843655046393531
966887567555247
10801049375646255
表2  不同结构元尺寸分割结果
图8  预测最佳拟合直线(蓝线)与实测直线(红线)关系
样方实测提取正确提取漏分错分精度
编号株数/株株数/株株数/株株数/株株数/株OA%CE%OE%AR%
11161009125984.09.025.091.0
26470622891.411.42.8688.6
3565846101296.620.717.279.3
4818371101297.614.512.085.5
5707160101198.615.514.184.5
664745861686.521.68.178.4
779857091592.917.610.682.4
84346376993.519.613.080.4
9666756101198.516.414.983.6
1080757010593.36.713.393.3
平均值71.972.962.19.710.693.2915.2513.1384.75
总体7197296219810898.6314.8113.4485.19
表3  样地株数精度分析表
1 Chen Yaning,Hao Xingming,Li Weihong,et al.An Analysis of the Ecological Securityand Ecological Water Requirements in the Inland River of Arid Region[J].Advances in Earth Science,2008,23(7):732-738.陈亚宁,郝兴明,李卫红,等.干旱区内陆河流域的生态安全与生态需水量研究[J].地球科学进展,2008,23(7):732-738.
2 Crowther T W,Glick H B,Covey K R,et al.Mapping Tree Density at a Global Scale[J].Nature,2015, 525(7568):201-205.
3 Qin Xianlin,Li Zengyuan,Yi Haoruo.Extraction Method of Tree Crown Using High-resolution Satellite Image[J].Remote Sensing Technology and Application,2005,20(2):228-232.覃先林,李增元,易浩若.高空间分辨率卫星遥感影像树冠信息提取方法研究[J].遥感技术与应用,2005,20(2):17-21.
4 Krizhevsky A,Sutskever I,Hinton G E.ImageNet Classification with Deep Convolutional Neural Networks[J].International Conference on Neural Information Processing Systems,2012,60(2):1097-1105.
5 Li W,Fu H,Yu L,et al.Deep Learning based Oil Palm Tree Detection and Counting for High-resolution Remote Sensing Images[J].Remote Sensing,2016,9(1):22-33.
6 Ronneberger O,Fischer P,Brox T.U-Net:Convolutional Networks for Biomedical Image Segmentation[C] //Medical Image Computing and Computer-Assisted Intervention —MICCAI,Munich,2015.
7 Hu Jianbo,Zhang Jian.Unmanned Aerial Vehicle Remote Sensing in Ecology:Advances and Prospects[J].Acta Ecologica Sinica,2018,38(1):20-30.胡健波,张健.无人机遥感在生态学中的应用进展[J].生态学报,2018,38(1):20-30.
8 Li Deren,Li Ming.Research Advance and Application Prospect of Unmanned Aerial Vehicle Remote Sensing System[J].Geomatics and Information Science of Wuhan University,2014,39(5) :505 – 513.李德仁,李明.无人机遥感系统的研究进展与应用前景[J].武汉大学学报·信息科学版,2014,39(5):505-513.
9 Wang Xiaoqin,Wang Miaomiao,Wang Shaoqiang,et al.Extraction of Vegetation Information from Visible Unmanned Aerial Vehicle Images[J].Transactions of the Chinese Society of Agricultural Engineering,2015,31(5):152-159.汪小钦,王苗苗,王绍强,等.基于可见光波段无人机遥感的植被信息提取[J].农业工程学报,2015,31(5):152-159.
10 Fu Wenjie,Lin Mingsen.Study on Extracting of Eoquat Information Using SVM and Gray-level Co-occurrence Matrix from QuickBird Image[J].Remote Sensing Technology and Application,2010,25(5): 695-699.傅文杰,林明森.利用SVM与灰度共生矩阵从QuickBird影像中提取枇杷信息[J].遥感技术与应用,2010,25(5):695-699.
11 Feng Jingjing, Zhang Xiaoli, Liu Huiling.Single Tree Crown Extraction based on Gray Gradient Image Segmentation[J].Journal of Beijing Forestry University,2017,39(3):16-23.冯静静,张晓丽,刘会玲.基于灰度梯度图像分割的单木树冠提取研究[J].北京林业大学学报,2017,39(3):16-23.
12 Culvenor D S.An Algorithm for The Delineation of Tree Crowns in High Spatial Resolution Remotely Sensed Imagery[J].Computers and Geosciences,2002,28(1):33-44.
13 Katoh M,Gougeon F A,Leckie D G.Application of High-resolution Airborne Data Using Individual Tree Crowns in Japanese Conifer Plantations[J].Journal of Forest Research,2009,14(1):10-19.
14 Gomes M F,Maillard P.Detection of Tree Crowns in Very High Spatial Resolution Images[J]. Environmental Applications of Remote Sensing,2016(8):41-71.
15 Pouliot D,King D.Approaches for Optimal Automated Individual Tree Crown Detection in Regenerating Coniferous Forests[J].Canadian Journal of Remote Sensing,2005,31(3):255-267.
16 Zhang Ning,Feng Yuewen,Zhang Xiaoli,et al.Extracting Individual Tree Crown by Combining Spectral and Texture Features from Aerial Images[J].Journal of Beijing Forestry University,2015,37(3):13-19.张凝,冯跃文,张晓丽,等.结合航空影像纹理和光谱特征的单木冠幅提取[J].北京林业大学学报,2015,37(3):13-19.
17 Shen Liqiang,Jiang Renrong,Wang Peifa.A Method for Individual Tree-crown Extraction from High Spatial Resolution Imagery[J].Remote Sensing Information,2017,32(3):142-148.沈利强,姜仁荣,王培法.一种高分辨率遥感图像单木树冠信息提取方法[J].遥感信息,2017,32(3):142-148.
18 Bu Fan,Shi Yuli.The Comparison of Urban Tree Crown Extraction based on Airborne LiDAR Elevation Difference and High Resolution Imagery[J].Remote Sensing Technology and Application.2017,32(5):875-882.
18 卜帆.石玉立,机载LiDAR 高差和高分影像的城市树冠提取比较[J].遥感技术与应用,2017,32(5) :875-882.
19 Liu Qingwang,Li Shiming,Li Zengyuan,et al. Review on the Applications of UAV-based LiDAR and Photogrammetry in Forestry[J].Scientia Silvae Sinicae,2017,53(7):134-148.
19 刘清旺,李世明,李增元,符利勇,胡凯龙.无人机激光雷达与摄影测量林业应用研究进展[J].林业科学,2017,53(7):134-148.
20 Zhou Honghua,Li Weihong,Sun Huilan.Reconstruction of Groundwater Depth Using Tree-rings of Populus Euphratica in The Lower Tarim River[J].Scientia Silvae Sinicae,2018,54(4):11-16.周洪华,李卫红,孙慧兰.基于胡杨年轮的塔里木河下游地下水埋深历史重建[J].林业科学,2018,54(4):11-16.
21 Long J,Shelhamer E,Darrell T. Fully Convolutional Networks for Semantic Segmentation[J].Computer Vision and Pattern Recognition,2015,79(11):3431-3440.
22 Dong Yunya,Zhang Qian. A Survey of Depth Semantic Feature Extraction of High-resolution Remote Sensing Images based on CNN[J]. Remote Sensing Technology and Application,2019,34(1):1-11.董蕴雅,张倩.基于CNN的高分遥感影像深度语义特征提取研究综述[J].遥感技术与应用,2019,34(1):1-11.
23 Wu Guangming,Chen Qi,Shibasaki Ryosuke, et al.High Precision Building Detection from Aerial Imagery Using a U-Net Like Convolutional Architecture[J]. Acta Geodaetica et Cartographica Sinica, 2018,47(6):864-872.伍广明,陈奇,Ryosuke Shibasaki,等.基于U型卷积神经网络的航空影像建筑物检测[J]. 测绘学报,2018,47(6):864-872.
24 Vincent L,Soille P.Watersheds in Digital Spaces:An Efficient Algorithm based on Immersion Simulations[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1991,13(6):583-598.
25 Gonzales R C,Woods R E,et al.Digital Image Processing [M].Beijing:Digital Image Processing.2011.冈萨雷斯,伍兹,等.数字图像处理[M].北京:电子工业出版社,2011.
26 Guo Yushan,Liu Qingsheng,Liu Gaohuan,et al.Individual Tree Crown Extraction of High Resolution Image based on Marker-controlled Watershed Segmentation Method[J]. Journal of Geo-information Science,2016,18(9):1259-1266.郭昱杉,刘庆生,刘高焕,等.基于标记控制分水岭分割方法的高分辨率遥感影像单木树冠提取[J].地球信息科学学报,2016,18(9):1259-1266.
27 HeYi,Zhou Xiaocheng,HuangHongyu,et al.Counting Tree Numberin Subtropical Forest Districts based on UAV Remote Sensing Images [J].Remote Sensing Technology and Application,2018,33(1):168-176.何艺,周小成,黄洪宇,等.基于无人机遥感的亚热带森林林分株数提取[J].遥感技术与应用,2018,33(1):168-176.
[1] 刘天福,陈学泓,董琪,曹鑫,陈晋. 深度学习在GlobeLand30-2010产品分类精度优化中应用研究[J]. 遥感技术与应用, 2019, 34(4): 685-693.
[2] 林志玮,丁启禄,黄嘉航,涂伟豪,胡典,刘金福. 基于DenseNet的无人机光学图像树种分类研究[J]. 遥感技术与应用, 2019, 34(4): 704-711.
[3] 王颖洁,张荞,张艳梅,蒙印,郭文. 基于深度卷积神经网络的油罐目标检测研究[J]. 遥感技术与应用, 2019, 34(4): 727-735.
[4] 王宇,杨艺,王宝山,王田,卜旭辉,王传云. 深度残差神经网络高分辨率遥感图像建筑物分割[J]. 遥感技术与应用, 2019, 34(4): 736-747.
[5] 徐梦竹, 徐佳, 邓鸿儒, 袁春琦. 基于全极化SAR影像的海岛地物分类[J]. 遥感技术与应用, 2019, 34(3): 647-654.
[6] 李淑敏, 冯权泷, 梁其椿, 张学庆. 基于深度学习的国产高分遥感影像飞机目标自动检测[J]. 遥感技术与应用, 2018, 33(6): 1095-1102.
[7] 汪航,师茁. 基于MODIS时间序列数据的春尺蠖虫害遥感监测方法研究—以新疆巴楚胡杨为例[J]. 遥感技术与应用, 2018, 33(4): 686-695.
[8] 闫鹏飞,明冬萍. 尺度自适应的高分辨率遥感影像分水岭分割方法[J]. 遥感技术与应用, 2018, 33(2): 321-330.
[9] 何艺,周小成,黄洪宇,许雪琴. 基于无人机遥感的亚热带森林林分株数提取[J]. 遥感技术与应用, 2018, 33(1): 168-176.
[10] 何海清,庞燕,陈晓勇. 面向遥感影像场景的深度卷积神经网络递归识别模型[J]. 遥感技术与应用, 2017, 32(6): 1078-1082.
[11] 卜帆,石玉立. 机载LiDAR高差和高分影像的城市树冠提取比较[J]. 遥感技术与应用, 2017, 32(5): 875-882.
[12] 周在明,杨燕明,陈本清. 基于无人机影像的滩涂入侵种互花米草植被信息提取与覆盖度研究[J]. 遥感技术与应用, 2017, 32(4): 714-720.
[13] 张正健,李爱农,边金虎,赵伟,南希,靳华安,谭剑波. 基于无人机影像可见光植被指数的若尔盖草地地上生物量估算研究[J]. 遥感技术与应用, 2016, 31(1): 51-62.
[14] 周艳飞,张绘芳,李霞,杨帆,丁程锋. 基于高分辨遥感数据的胡杨与柽柳树冠提取[J]. 遥感技术与应用, 2015, 30(3): 510-517.
[15] 李政,李永树,唐敏. 一种求解无人机影像外方位角元素的方法[J]. 遥感技术与应用, 2014, 29(3): 506-510.