专论 网络优先出版 版本 2
下载
无人机数字摄影测量与激光雷达在地形地貌与地表覆盖研究中的应用及比较
Applications of UAV digital aerial photogrammetry and LiDAR in geomorphology and land cover research
: 2018 - 08 - 11
: 2018 - 10 - 16
: 2018 - 10 - 25
122 0 0
摘要&关键词
摘要:无人机数字摄影测量(UAV-DAP)与激光雷达(LiDAR)凭借其机动灵活、高效便捷和高分辨率的特点,在地形地貌、生态监测、工程勘察、环境规划、林业资源清查等领域得到广泛应用。本文针对数字摄影测量与激光雷达的技术特征和应用趋势,着重比较两者在数据采集、数据处理、应用领域以及成本耗费等方面的区别,分析了两种技术在林业资源清查、地形地貌研究、灾害防控等领域的最新应用动态,并且基于两者的技术特点和发展动态提出了进一步的可能应用前景。论文指出,基于运动结构重建算法的数字摄影测量技术获得的数字地表模型在一定条件下可以达到激光雷达技术的超高空间分辨率程度(如0.2 m×0.2 m),但是数字摄影无法穿透植被冠层,而激光雷达可以较好地穿透植被层从而获取植被及地表信息。然而数字摄影测量技术设备简单、操作方便,成本低廉,并具有较高的空间分辨率,因而能够和高精度、高耗费、大数据量的激光雷达技术形成优势互补。无人机数字摄影测量与激光雷达技术是林业资源清查、地形地貌研究、灾害防控等领域在快速响应、高精度调查、多时期扫描等方面进一步发展的重要突破口。
关键词:无人机,数字摄影测量,激光雷达,地形地貌, 地表覆盖
Abstract & Keywords
Abstract: Background, aim and scope. The unmanned aerial vehicle (UAV) digital aerial photogrammetry (DAP) and aerial lidar scanning (ALS) have been widely applied in geomorphology research, ecosystem monitoring, engineering investigation, environmental planning, and forest inventory due to their advantages of high flexibility, high efficiency, and high resolution. Photogrammetry is the science of measuring ranges from photographs, especially for recovering the space positions of optical surface points. Photogrammetry can be dated back to mid-19th century when it was the beginning of modern photography. LiDAR (Light Detection and Ranging) is a ranging technique that measures distance to a target by illuminating the target with pulsed laser light and measuring the reflected pulse with a sensor. At the end of 20th century, photogrammetry was caught up by LiDAR because of its series of shortcomings, such as heavy equipment, low efficiency, and high expense. It did not take long before LiDAR went beyond photogrammetry in many applications. However, photogrammetry came back to the stage due to the fast development of small UAV, digital imaging devices, computational advance, photogrammetry algorithm, and related software development during the past decade. In this study, we aim to compare DAP and ALS and discuss their future trends. Materials and methods. This paper reviews current major advantages, applications and perspectives of UAV DAP and ALS. We briefly analyzed the fundamental techniques and principles of DAP and ALS through typical research and study cases. We focused on the differences between these two technologies in data acquisition, data processing, flight planning, cost, advantages, and applications through synthesizing published scientific results as well as practical operational considerations. Results. The UAV remote sensing technologies, including digital aerial photogrammetry, aerial lidar scanning, and high spectrum imaging, have provided a flexible platform for terrain- and vegetation-based surface observations. The resolution of DAP can be equal to that of ALS, and the former is much more flexible and economical. Discussion. ALS needs much more complex facilities than DAP to launch an aerial survey, which is difficult and expensive to operate and mostly contracted out to professional companies. In contrast, DAP is rather easy, for a small UAV with a digital camera can carry out an entire aerial survey in short time. The data processing of ALS is “direct” due to the raw outcome of ALS is point cloud. While DAP needs to extract point clouds from aligned images in the first place, the whole process is rather efficient owing to the SfM (Structure from Motion) algorithm based professional software, which mostly have applied the CUDA (Compute Unified Device Architecture) techniques to accelerate the whole processing. DAP and ALS have been applied to forest inventory, geomorphology evolution, glacier change, gully erosion, and many other fields, and the resolution of derived DSM (Digital Surface Model), DEM (Digital Terrain Model), CHM (Canopy Height Model) can be comparable between DAP and ALS, though ALS has more advantages in high resolution research. Conclusions. UAV-based DAP and ALS technologies have four features: quick response, quick deployment, quick result, and high resolution. DAP and ALS complement each other in topography study, landform research, gully erosion, glacier change, forest inventory, and ecosystem survey. These two technologies plus high spectrum imaging offer significant complement to earth surface observations in satellite-based remote sensing, which often has limitations in spatial and temporal resolutions. The resolution of SfM-based DAP results can be as high as 0.5m×0.5m and even 0.2m×0.2m, which makes DAP competitive to ALS. But DAP cannot take the place of ALS as lidar can penetrate tree canopy and retrieve point clouds of terrain surface and subjects above the surface, such as trees and buildings. DAP must be carried out in a bright environment, which means a sunny day, while ALS aerial survey can be conducted in a cloudy day. The SfM-based DAP requires photos having enough overlapping areas, 30% to 50%, to ensure right alignment. However, DAP has a vital advantage over ALS in terms of cost (the cost of the former can be one third of the latter). A small UAV (such as DJI phantom 4) with digital camera, which is only 1280 grams, can carry out a DAP aerial survey in a short time, whereas a lidar sensor can be twice the weight of a DJI phantom 4. It is quicker and easier for DAP to operate in scenario response, aerial deployment, and result presentation than those of ALS. DAP technology can complement ALS in geomorphology research in the Loess Plateau, and the former can take the place under certain circumstances. DAP can acquire rather “real” terrain surface data in the Loess Plateau during winter and early spring when slopes and gullies are covered with sparse trees. With the help of historical high-resolution terrain data (lidar-derived DEM), the DAP results can be more accurate when generating DSM, DEM, and CHM. Recommendations and perspectives. DAP and ALS have competed over years. These two technologies have contributed revolutionary changes in observation and quantification of the Earth’s surface study. Consequently, DAP and ALS offer tremendous opportunities in complementary ways in forest inventory, ecosystem survey, geomorphology invetigation, and land cover research.
Keywords: UAV; digital photogrammetry; LiDAR; geomorphology; land cover
摄影测量学(photogrammetry)是通过物体表面点不同角度的相片来还原点的空间位置,然后利用足够多的空间点重构物体形状的一门信息科学。摄影测量学最早可以追溯到19世纪中期,与照相术(photography)属同一时代。最早的摄影测量学是通过照片里的比例物来确定物体大小或者距离。随着两次世界大战,摄影测量学得到了飞速发展,经历了模拟摄影测量、解析摄影测量、数字摄影测量三个阶段。传统的摄影测量学是利用光学摄影机获取的影像还原被摄物体的形状、大小、性质和相互关系,应用于建筑物、地形测量、工程勘察、生态资源规划等(张祖勋等,2000)。今天,随着计算机技术、成像设备、摄影测量算法、小型无人机技术的迅猛进步,立体摄影测量学(Stereophotogrammetry)得以蓬勃发展。从室内三维建模到古建筑模型重建,从作物产量监测到森林资源清查,从沟道侵蚀研究到冰川形态监测,立体摄影测量学在不同空间尺度的3D建模领域正在发挥巨大作用(Aguilar et al,2017;Frankl et al,2015;Jin et al,2017;Neugirg et al,2016;Santoso et al,2016;Turner et al,2014;White et al,2013)。
20世纪末,在立体摄影技术发展的同时,激光雷达(LiDAR, Light Detection and Ranging,激光探测与测量技术)快速跟进,并呈现出超越前者的发展势头(Baltsavias,1999)。在地学领域,地表观测和定量分析在激光雷达技术提供的极高空间分辨率数据的支持下正在进行着一场变革式的发展(Harpold et al,2015)。为了区分摄影测量与激光雷达技术在地形与地表覆盖研究中各自的优势,使得科技工作者能够在科学要求与经费支持的平衡中找到合适的技术方案,本文拟从激光雷达和摄影测量成像各自的技术原理入手,对数据获取、数据处理、航测规划、应用实例等方面进行比较,最后总结两者的优势和发展趋势,从而为当前地形地貌与地表覆盖物研究提供更加广阔的研究思路与高效优质的技术支持。
1   摄影测量
在20世纪末,国内摄影测量相比于国外起步较晚,但在中国国家地理信息测绘局、各勘探院所、众多高校以及广大科研工作者的全力推动之下,依然在短时间内取得了长足的进步,在包括摄影测量、遥感、地理信息系统等领域取得了一系列成果和突破(李德仁等,2001)。基于影像的3D建模(Image-Based 3D Modeling)是立体摄影测量学的一个重要类别,其原理是通过不同角度的相片来还原点的空间位置(图1a)。现今主要依据运动结构重建方法(SfM, Structure from Motion)对具有时间序列的多角度、多位置、有重叠的影像进行空间点位置解算和提取(图1b),进而得到空间点云以重建物体的3D模型(Bemis et al,2014)。在当前研究中,立体摄影测量学有许多名称,如基于图像建模(image-based modeling)(Frankl et al,2015),数字航空摄影测量(DAP,digital aerial photogrammetry)(White et al,2016),运动多视点立体结构方法(structure-from-motion multi view stereo approach)(Piermattei et al,2015)等等;但总体来说,这些方法的操作原理基本相同,数据获取方法大体相似,结果处理过程存在可比性。


图1   摄影测量原理示意图
Fig.1 Schematic diagram of photogrammetry
a早期摄影测量学基本原理;b基于SfM的时序影像3D建模;修改(Bemis et al,2014)a Principle of the early photogrammetry; b 3D image modeling based on SfM(Bemis et al, 2014)
早期的模拟摄影测量和解析摄影测量往往利用载人飞机进行较大范围的航空摄影,然后将获得的航空相片利用相应的立体成像仪人工进行地图绘制,在人力物力上耗费较多。到了20世纪末期,航空摄影取得了长足的进步,发展到数字摄影测量,在内业制图上用计算机取代了人工,大大提高了制图效率,但是数字化程度依然不够;如果没有影像自动处理算法和理论的突破,就很难得到进一步的发展(李德仁,2000)。随着小型无人机和激光探测设备的快速发展,便捷高效的激光雷达逐渐抢占航空摄影测量的地位(Baltsavias,1999),在地形地貌、土壤侵蚀、冰川变化、雪被遥感、林业调查、生物多样性等领域迅速获得广泛应用(Cavalli et al,2013;Harpold et al,2014;Harpold et al,2015;Laurin et al,2014;Neugirg et al,2016;Shellberg et al,2016;Tarolli,2014)。但是随着数码成像以及计算机技术的快速发展,摄影测量在21世纪焕发了新的生机(White et al,2013),从模拟摄影测量发展到新时代的数字摄影测量技术。基于影像的数字摄影测量三维重建要求影像具有较高的重叠率、便于识别的特征点、合适的图像尺寸以及空间分辨率(狄颖辰等,2011;朱锋等,2014)。当前,低成本小型无人机、高像素数码相机以及基于计算机视觉的三维重建算法的快速发展,使得数字航空摄影测量成为一种低成本的地形地貌、森林植被结构测量的选择方案(Dandois and Ellis,2013;刘清旺等,2017)。
常用的基于影像的三维建模方法包括运动结构重建(SfM,Structure from Motion)和半全局匹配(SGM,Semi-Global Matching)算法等。SfM算法抛弃了全球卫星导航系统和惯性制导数据,直接将摄影测量与计算机视觉算法结合,对不同距离、不同视角的重叠影像自动提取影像特征并进行匹配(Dandois and Ellis,2010;Ota et al,2015;Snavely et al,2008;St-Onge et al,2015;Turner et al,2012)。SGM方法用代价函数约束像元匹配的概率,采用全方向路径优化提高匹配效率(Gehrke et al,2012;Hirschmuller,2005;Penner et al,2015;Stepper et al,2015;White et al,2015)。典型的基于影像的三维建模软件参见表1。
 
表1 典型基于影像的三维建模软件
Tab.1 Typical image-based 3D modeling software
软件名称算法官方网址商业/开源
Software nameAlgorithmOfficial websitesBusiness/open source
Pix4DSfMhttps://pix4d.com & https://pix4d.com.cn/商业软件 business
Agisoft PhotoscanSfMhttp://www.agisoft.com & http://www.agisoft.cn商业软件 business
LiMapperSfMhttp://www.lidar360.com/商业软件 business
BundlerSfMhttp://www.cs.cornell.edu/~snavely/bundler/开源 open source/GNU
EcosynthSfMhttp://ecosynth.org/开源 open source
COLMAPSfMhttps://demuc.de/colmap/开源 open source/GNU
VisualSFMSfMhttp://ccwu.me/vsfm/开源 open source
注:上述软件大多采用了基于NVIDIA显卡的CUDA技术以进一步提高运算效率。
Note: Most of the software above use the NVIDIA GPU-based CUDA technology to increase computational efficiency.
2   激光雷达
激光雷达技术凭借其极高的空间分辨率和灵活的观测范围助力地表观测与定量化研究变革性发展,根据激光雷达点云衍生的各种数据已经在地形学、水文学以及生态学等多领域产生了一系列重大成果(Harpold et al,2015)。LiDAR技术依赖不同的平台,包括地基激光扫描(TLS,terrestrial laser scanning),机载激光雷达扫描(ALS,airborne laser scanning),星载卫星激光扫描(SLS,spaceborne laser scanning)等多种平台(Glennie et al,2013;Stennett,2004),结合其它的观测数据,在地形地貌变化、雪被遥感、冰川变化、植被信息、生物量监测等方面发挥了重要作用(Cavalli et al,2013;Harpold et al,2014;Harpold et al,2015;Laurin et al,2014;Tarolli,2014)。
机载激光雷达扫描技术(ALS)凭借其灵活可靠、范围广、易用性高等优点,在激光雷达技术的多种应用分支中拔得头筹。一套基本的ALS系统包括激光测距装置、光机扫描仪、控制-监测-记录装置、动态差分GPS接收机、姿态测量装置IMU(如图2a)(Wehr and Lohr,1999)。应用于林业的ALS设备,其激光测距装置通过扫描仪能向地面每秒发射50000-150000束激光脉冲甚至更多,而且能够接收到每个射出的激光脉冲的至多5个返回信号。ALS沿飞行方向前进,扫描仪则不断进行垂直前进方向上的脉冲发射-接收过程(扫描带宽),得到距离(z)和位置(x,y)信息,从而构建出一系列三维空间点云,并且这些扫描带宽之间会有50%左右的重叠区以保证足够密集以及均一的点云密度(Reutebuch et al,2003;White et al,2013)。ALS对于平原以及丘陵地形可以获得很好的空间点云,但是在较大高差地形环境下(如陡崖)点云密度迅速减少,数据质量会出现不同程度的降低;而且浓密森林冠层条件下的空间点云信息也会出现一定程度的误差(Gatziolis et al,2010;Tinkham et al,2012)。ALS获得的三维空间点云数据结合地面站数据校正,然后经过数据处理软件进行点云抽稀,分离成地面点和非地面点(如树木、建筑等),从而获得地表植被及地形垂向信息,地面点用于构建高精度的数字高程模型(DEM),非地面点则用于构建高精度的地表数字模型(DSM)以及植被冠层高度模型(CHM)等,其空间分辨率可以达到1×1 m,图2c、d、e是郭庆华等(2014)对 1 m航片、30 m DEM和1 m激光雷达数据衍生的DEM进行对比,可以发现激光雷达获取的地形数据在空间分辨率方面具有较大优势。


图2   机载激光雷达设备及观测结果示意图
Fig.2 ALS equipment and observations
a ALS设备构成及工作原理示意图(Wehr and Lohr,1999);b 八旋翼小型无人机UAV-ALS设备(Guo et al,2017); c 航空摄影拍摄的1m分辨率航片;d 美国地质勘探局(USGS)生成的30 m分辨率DEM数据;e 机载激光雷达生成的1m分辨率DEM数据(郭庆华等,2014)a Basic frame and principles of ALS(Wehr and Lohr, 1999); b New 8-rotor UAV-ALS(Guo et al, 2017);c DAP image of 1×1 m; d DEM of 30×30 m from USGS; e ALS DEM of 1×1 m(郭庆华等,2014)
经过近20年的发展,早期的大型ALS载人飞行平台已经发展成小型无人机平台(如图2b),其适用范围更广,而且在安全性、灵活性、经济型上优于早期设备(Guo et al,2017),目前在国内的林业资源调查、铁路巡线、地形勘测等领域已经得到应用。国外在ALS上已经有十几年的发展,具有技术和理论优势;而我国科学界依托逐渐强大的综合国力已经在包括ALS、TLS的多个激光雷达开发应用领域取得了长足的进步,郭庆华、刘清旺、于景鑫等利用小型无人机-激光雷达技术在森林生态系统监测、生物多样性监测等领域取得了一系列成果(郭庆华等,2016a;郭庆华等,2014;郭庆华等,2016b;刘清旺等,2017;于景鑫等,2015)。
3   数字航空摄影测量与机载激光雷达技术对比
White等(2013)在利用数字航空摄影测量(DAP)和机载激光雷达(ALS)两种方法进行森林资源清查,比较两种方法获得的点云、数字地表模型、冠层高度模型的精度,最后发现摄影测量方法所得模型的空间精度和激光雷达所得结果的空间精度基本一致。即使当前激光雷达技术已经无人机化、小型化,由于激光探测设备价格昂贵,其总体使用成本较高,限制了激光雷达技术的更广泛应用;因此寻找一种能在价格和空间分辨率间平衡的替代技术就成为了一个需要思考的问题。以下将从数据获取、数据处理、产品等方面来进行两者的对比。
3.1   数据获取
3.1.1   数据采集设备
与激光雷达的TLS、ALS等多种搭载平台类似,数字航空摄影测量也可以使用地基或者机载平台(Bemis et al,2014),但是摄影测量的影像采集设备更加轻便,一架大疆精灵3小型无人机和高像素数码相机(总重量约为1.3kg)就可以达到航测的基本要求(Ajayi et al,2017)。相比之下,机载激光雷达扫描由于需要动态差分GPS接收机、姿态测量装置(IMU)、激光扫描测距系统、信号接收与发送系统等(图2a),单独测量部分总重量就达到3kg左右,如果加上八旋翼无人机,总重可能达到8kg。
3.1.2   数据处理
机载激光雷达能直接获得植被及地表点云数据,通过软件(如LiDAR360,TerraScan,LP360等)进行地表点和非地表点的划分,从而获得高分辨率的植被和地面高程信息。数字航空摄影测量对具有空间坐标和镜头信息的多角度、多位置、有重叠的影像(照片)进行空间点位置解算和提取,生成基于影像的点云,然后实现三维模型重建。与激光雷达相比,数字航空摄影测量有两点需要注意:(1)航拍影像重叠度在航线上要求60%或以上,在航线间要求40%左右;(2)在较强光照条件(晴天)下进行航测(White et al,2013)当前主流基于SfM方法的摄影测量三维建模软件如Agisoft PhotoScan、Pix4D、LiMapper等(图3a),远比传统的立体摄影测量方法高效。本文使用大疆精灵4小型无人机在延安宝塔区顾屯村拍摄了144张照片,然后用PhotoScan软件建立三维地形,结果如图3b。


图3   基于影像的三维重建软件及处理结果
Fig.3 Image-based 3D reconstruction software and processing results
需要指出的是,上述软件除了使用SfM算法,还大多支持基于英伟达计算显卡的通用并行计算架构技术(CUDA,Compute Unified Device Architecture),能够依靠英伟达显卡成百上千的流处理单元进行图像拼接的并行高速处理,相比于没有CUDA支持的情形其速度能够提升5-10倍。
数字航空摄影测量所得重叠影像进行拼接的关键是基于影像中地物的反射特性和几何形状的相似性找到相关点,然后通过区域匹配或者地物匹配进行影像拼接。区域匹配是通过一个窗口或者多个影像像元的对比来寻找最佳匹配影像,地物匹配方法则依赖于地物实体对象,如点、线、多边形等。需要注意的是,除了影像拼接,参数的立体匹配、影像分辨率、太阳入射角、观察几何都会对基于影像建模的摄影测量成像产生显著的影响,进而影响生成DSM的精度和分辨率(St-Onge et al,2008)。
3.1.3   航测与费用
在制定航线时,DAP在沿航线方向和垂直航线方向有更宽的割幅(swath)和重叠区,为影像拼接提供充足的数据冗余以便于计算。ALS受限于激光扫描仪的能量强度,对于陡峭地形的观测存在不足。而无论上述哪种方法,增加控制点都能够有效地提高精度(Baltsavias,1999;刘清旺等,2017)。DAP可以比ALS飞的更高、更快,在相同时间内可以选择拍摄更大的区域或者增加重叠区域。这需要依据实际无人机状况而定。但是由于DAP是基于影像建模,因此光照对其影响很大,需要在晴天下进行。相比之下,ALS受光照影响较小能在阴天进行航测,但实际操作中也多在晴天下进行。
DAP的价格是机载激光雷达技术的三分之一甚至更低(White et al,2013)。现在,小型无人机搭载高像素数码相机基本能够满足DAP需求,Ajayi等(2017)采用大疆精灵3无人机(搭载SONY 1200万像素数码相机)进行小区域地形监测,飞行时间30分钟,控制距离4km,设备总价格为1-2万。而采购ALS设备,其基本包括八旋翼无人机(20-30万)和LiDAR探测系统(50-60万);如果选择外包给相应公司则根据航测区域面积收费。
3.2   优势比较
ALS通过软件(如TerraScan,LP360,LiDAR360等)进行地表点和非地表点的划分获得点云,其运行速度快于基于影像获取点云方法(Leberl et al,2010)。但是新算法和高效计算设备的发展使得DAP的运行效率大大提升,ALS的效率优势逐渐丧失(White et al,2013)。综合上述,可以得到航空摄影测量技术与机载激光雷带技术的优势比较,见表2。
 
表2 数字航空摄影测量技术和机载激光雷达技术比较
Tab.2 Comparison of advantages between DAP and ALS
参数优势方备注参考
ParameterAdvantagesNotesReference
数据获取Acquisition
飞行规划Flight planning数字摄影测量DAP数字影像在航线和垂直航线方向上要求较高的重合度;ALS受限于激光能量和自身飞行效能因而在陡峭地形操作困难。两者都需要充分考虑到利用地面控制点提高准确度。DAP has wide swaths and overlap in both the along-track and cross-track directions. In steep terrain, planning for ALS is particularly challenging as maximum flying height is restricted by laser power. The number and positioning of GPS base stations must be considered.Baltsavias, 1999
航测时间和区域Survey time and area数字摄影测量DAP摄影测量的机载平台可以比激光雷达的飞行速度更快,高度更高。摄影测量的视场角(≤75°)比ALS的视场角(≤25°)开阔,因此前者在相同的飞行时间可以获得更大区域的信息。Image platforms are able to fly higher and faster than ALS platforms. Imaging instruments will typically have a FOV of 75°; ALS FOV (for forest applications) are ≤25°. Thus, for the same number of flying hours, image acquisition can cover a much larger area.Baltsavias, 1999 Leberl et al, 2010
飞行条件Flying conditions机载激光雷达ALS机载激光雷达设备巡航时间更长,但是不适用与阴影区。摄影测量则会受到光照和拍摄角度的强烈影响。More flying hours per day are possible with ALS, which is insensitive to shadow. Imagery is strongly influenced by solar illumination and view angles (sun, surface, and sensor geometry).Baltsavias, 1999Gehrke et al, 2011
数据处理Processing
效率Efficiency机载激光雷达ALS机载激光雷达从航测到生成点云的时间比摄影测量方法略短,但是两者之间的效率差距随着技术的发展在逐渐缩小。Time required to go from acquisition to point cloud is shorter for ALS, although this advantage has narrowed over time with the advent of fully digital photogrammetric workflows.Baltsavias, 1999Leberl et al, 2010
数据产品Products
产品Outputs数字摄影测量/机载激光雷达DAP/ALS摄影测量只能获得数字地表模型,而激光雷达则可以穿透植被层进而获得DSM、DEM以及CHM。但是摄影测量在获得DSM的同时能够得到地面影像,便于地面物体解译;而ALS则需要另外加载光谱摄像设才能够实现相同功能。Basically, DAP obtains DSM, and ALS could penetrate canopy to get DSM, DEM, and CHM. However, DAP can obtain both DSM and surface image, facilitating the interpretation of surface targets. While ALS acquires additional image sensors to accomplish that.Baltsavias, 1999Bohlin et al, 2012Järnstedt et al, 2012
空间分辨率Resolution数字摄影测量DAP在相同花费下,摄影测量相对于ALS可以获得更高的点云密度,而且前者能够进行更大范围的遥感采集或者增大影像重合度以提高解译精度。Much greater point densities are attainable with imagery than with ALS for a given cost, owing to smaller GSDs and image overlap that enables a large number of independent three-dimensional pixel matches. The higher density of points may enable a better representation of discontinuities.Leberl et al, 2010Baltsavias, 1999
准确度Accuracy机载激光雷达ALS摄影测量和激光雷达对于给定区域物体的观测准确度相差不大。对于垂向测量,ALS的准确度要高于摄影测量;而且即使提高点云密度也不一定能够提高摄影测量的垂向观测准确度。Accuracies are comparable between ALS and imagery for well-defined surfaces. More direct comparisons of canopy heights and other inventory metrics derived from the two data sources are required in a range of forest environments. The higher density of points in image-based point clouds does not necessarily result in greater vertical accuraciesHaala et al, 2010Leberl et al, 2010
3.3   数字航空摄影测量与机载激光雷达技术应用比较
3.3.1   林业资源勘察
数字航空摄影测量和机载激光雷达技术所得空间点云精度差距不大,都能达到0.5×0.5 m(White et al,2013),甚至更高的空间分辨率;但是前者是通过影像提取点云,而后者则是直接获取空间点云。White等(2013)利用两种技术获得森林树冠高度模型(两种点云的空间分辨率都是0.5×0.5 m),他们划分了一个61 m的横断面后发现,摄影测量的冠层高度模型横断面上不能将树冠之间的空隙进行区分,但是通过两者的树冠高度的对比拟合后发现在树冠部分两者测量的冠层高度能够比较好的吻合(图4a)。White等(2015)对比激光雷达和摄影测量方法获得的树冠点云模型的树高后发现,两者的相关系数能够达到0.9(图4b)。Goodbody等(2017)收集2013、2014、2015等年份的同一地区森林ALS和DAP点云数据,建立冠层高度模型后进行拟合分析,发现两者的数据精度相近,相关系数达到了0.98。国内,中国科学院植物研究所郭庆华等(2016b)采用小型无人机机载激光雷达技术进行林业资源勘察,进行新型激光雷达和数字航空摄影技术在林业生态方面的多层次对比研究(图4c),认为无人机遥感(包括激光雷达和摄影测量)可以很好的弥补地面监测与航天、卫星遥感之间的尺度空缺,使得地面监测结果更为准确,可更好地服务于局地到区域尺度的决策分析。


图4   机载激光雷达和摄影测量在林业资源清查中的应用比较
Fig.4 Comparison of ALS and DAP in forest inventory
a 机载激光雷达和基于影像建模获取树冠高度模型对比;修改(White et al,2013);b ALS和DAP方法获得树冠点云模型对比;修改(White et al,2015);c 无人机激光雷达和摄影测量获得同一片区域森林影像对比;修改(郭庆华等,2016b)a Comparison of CHM derived from ALS and DAP (White et al, 2013);b Comparison of canopy point cloud from ALS and DAP (White et al, 2015);c Comparison of forest image from ALS and DAP (郭庆华等,2016b);
3.3.2   地形地貌
激光雷达技术和数字摄影测量技术已经在国内外的地形地貌变化、土壤侵蚀、冰川研究中得到较大应用,而且其发展势头迅猛,逐渐成为联结地基地形地貌测量和天基卫星遥感的重要纽带。Neugirg等(2016)通过地基激光累到扫描和摄影测量方法研究地面沟道侵蚀(图5 a,b),认为激光雷达可以提供更为精细的地表量化信息,而基于运动结构重建方法的摄影测量技术对于整个集水区的地形变化则是较为理想的研究手段。Christian和Davis(2016)利用基于SfM方法的数字航空摄影测量方法研究山坡沟道形态,用PhotoScan软件处理航拍获得的时序影像后建立3D模型(DSM),然后将影像作为“贴图”附着在高程模型上从而重建出沟道形态(图5 f),他们将基于摄影测量获得的沟道DSM和基于LiDAR技术获得的0.5m分辨率DSM以及差值到1 m分辨率的地形数据进行对比,从中可以看出基SfM方法的DAP的DSM分辨率高于0.5 m,超过了LiDAR技术(图5 c,d,e)。


图5   LiDAR、DAP、DEM衍生地形数据比较
Fig.5 Comparison of topography data derived from LiDAR, DAP, and DEM
a TLS(地基激光雷达扫描)获取山坡点云;b 基于SfM方法的DAP得到的山坡点云c SfM-DAP获得的DSM;d 基于LiDAR技术的0.5 m分辨率DSM;e 空间差值为1m分辨率的DSM;f 利用PhotoScan重建DAP影像所得沟道3D正射模型(Christian and Davis,2016;Neugirg et al,2016)a Point cloud form TLS; b Point cloud from SfM-based DAP; c DSM from SfM-based DAP; d 0.5×0.5 m DSM from LiDAR; e 0.5×0.5 m DSM by spatial interpolation; f 3D modeling of a valley by DAP image via Photoscan(Christian and Davis, 2016;Neugirg et al, 2016)
Piermattei等(2015)利用基于SfM算法的地基摄影测量方法监测阿尔卑斯山一个小型冰川的质量变化,并且用地基激光扫描结果进行验证,两种方法得到的地表模型都是0.2×0.2 m空间分辨率,从对比结果可以看出摄影测量成像可以较好地补充激光扫描结果。张盈松等(2012)采用地面多基线数字摄影测量方法获取了中国天山东段博格达峰南坡的黑沟8号冰川末端影像,处理所得DEM数据在高程上的平均误差为1.92 m,标准偏差为3.47 m。以上研究表面数字摄影测量在典型冰川测量中能够发挥重要作用。需要注意的是立体摄影测量方法对于雪被不能达到理想效果,因为白雪近乎完全反射,在数码相机拍摄下无法获得特征点,因此雪被区的地形在摄影测量方法下完全没有模拟出来。而激光雷达技术由于能穿透一定深度的雪被,可以同时获取雪被和其下地表信息(依靠相关软件的解算与分离),因而能够很好地获得雪被区地形。
3.3.3   灾害防控
在2008年汶川地震过程中,由武汉大学领导研制的全数字摄影测量并行处理系统在汶川抗震救灾的应急航空摄影、海量应急航空影像数据快速处理、灾区概况实时监测等方面做出了重要贡献(张祖勋等,2009)。之后,无人机航空摄影测量技术逐渐成为监测、防控自然灾害的研究主流。彭大雷等(2017)使用无人机低空摄影技术调查甘肃黑方台地区36km2的黄土滑坡状况,并且通过历史积累数据对比滑坡前正射影像和滑坡后正射影像,根据滑坡发育特征将黄土滑坡分为4类:黄土-基岩型、浅层崩塌型、黄土-泥流型、静态液化型。认为航空(低空)数字摄影测量技术是对传统地质调查方法的扩展与补充,在提高工作效率的同时提高了地质调查的精度,使得以往一些几乎无法获得的数据成为可能。周明和邱凌云(2018)提出基于无人机的人际协同操作,以应对数量多、地形险峻、环境恶劣、防护措施缺乏的线路工程边坡的病害巡查。王帅永等(2016)利用无人机低空摄影系统获取老虎嘴滑坡区域的遥感影像、DEM和DOM(数字正射影像),并且重建三维空间场景,实现了对于强震区地质灾害的精细调查。张祖勋和吴媛(2015)提出了新时代下摄影测量的信息化与智能化特点,各类摄影测量传感器极大地促进了摄影测量技术的发展,扩展了摄影测量服务的内容,强调智能化的低空、无人机摄影测量是当前摄影测量技术发展的大趋势,而影像高重叠度和密集点云则是今年来摄影测量信息化发展的重要标志。
4   总结
无人机数字摄影测量和激光雷达技术在地形地貌变化、沟道侵蚀、冰川变化、植被信息、森林资源清查等领域互为补充,凭借其快速响应、快速部署、快速出图和高精度的特点弥补了地面常规监测和卫星遥感在时空尺度上的不足,研究者据此可以在实际需求和预算成本之间博弈出最佳选择。
(1)摄影测量与激光雷达测量精度相当,两者可以优势互补。基于SfM算法的数字航空摄影测量所得地表数字模型可以达到激光雷达的测量精度,通过合理布设地面控制点,两种技术的测量精度可以进一步提高。对于地形变化较小的区域,用激光雷达获取高精度地形,然后用摄影测量技术进行多期航测,从而获取不同季节的地表信息、植被冠层高度信息,进而实现低耗费的长序列多期观测。
(2)摄影测量操作简单且价格便宜。摄影测量依靠小型无人机和高像素数码相机就能进行数据采集,具有快速、便捷、成本低等特点。但是对于大区域的高精度观测,由于需要架设差分GPS设置地面控制点、使用长续航的大型无人机,依然需要专业航测团队的技术支持。
(3)摄影测量技术不能完全取代激光雷达。摄影测量只能得到数字地表模型,激光雷达则可以进行地面点和非地面点的区分,从而得到不同层次的数据,包括数字地表模型、数字高程模型、数字地形模型、冠层高度模型等。摄影测量受环境光照影响较为明显,在进行拍摄时要求影像重叠度在60%及以上,而且地物区分要比较明显。激光雷达技术具有更强的适应性,能够在弱光下进行数据采集,且不受地物色彩的影响。
致谢
感谢中国科学院植物研究所庞树鑫提供的信息和技术支持;感谢北京数字绿土科技有限公司提供的人员技术支持,感谢孙达、邓怀海、秦文卿在延安顾屯实验站的航空摄影测量和影像解析中提供的帮助。
致谢
Aguilar F J, Fernandez I, Casanova J A, et al. 2017. 3D coastal monitoring from very dense UAV-based photogrammetric point clouds//Eynard B, Nigrelli V, Oliveri S M, et al. Advances on mechanics, design engineering and manufacturing[M]. Switzerland: Springer, Cham, 881-888
Ajayi O G, Salubi A A, Angbas A F, et al. 2017. Generation of accurate digital elevation models from UAV acquired low percentage overlapping images[J]. International Journal of Remote Sensing, 38(8-10): 3113-3134.
Baltsavias E P. 1999. A comparison between photogrammetry and laser scanning[J]. Isprs Journal of Photogrammetry and Remote Sensing, 54(2-3): 83-94.
Bemis S P, Micklethwaite S, Turner D, et al. 2014. Ground-based and UAV-based photogrammetry: a multi-scale, high-resolution mapping tool for structural geology and paleoseismology[J]. Journal of Structural Geology, 69: 163-178.
Bohlin J, Wallerman J, Fransson J E S. 2012. Forest variable estimation using photogrammetric matching of digital aerial images in combination with a high-resolution DEM[J]. Scandinavian Journal of Forest Research, 27(7): 692-699.
Cavalli M, Trevisani S, Comiti F, et al. 2013. Geomorphometric assessment of spatial sediment connectivity in small Alpine catchments[J]. Geomorphology, 188: 31-41.
Christian P, Davis J. 2016. Hillslope gully photogeomorphology using structure-from-motion[J]. Zeitschrift Fur Geomorphologie, 60: 59-78.
Dandois J P, Ellis E C. 2010. Remote sensing of vegetation structure using computer vision[J]. Remote Sensing, 2(4): 1157-1176.
Dandois J P, Ellis E C. 2013. High spatial resolution three-dimensional mapping of vegetation spectral dynamics using computer vision[J]. Remote Sensing of Environment, 136: 259-276.
Frankl A, Stal C, Abraha A, et al. 2015. Detailed recording of gully morphology in 3D through image-based modelling[J]. Catena, 127: 92-101.
Gatziolis D, Fried J S, Monleon V S. 2010. Challenges to estimating tree height via LiDAR in closed-canopy forests: a parable from Western Oregon[J]. Forest Science, 56(2): 139-155.
Gehrke S, Downey M, Uebbing R, et al. 2012. A multi-sensor approach to semi-global matching//Shortis M, Paparoditis N, Mallet C. Xxii Isprs Congress, Technical Commission Iii[M]. Melbourne: ISPRS, 17-22
Glennie C L, Carter W E, Shrestha R L, et al. 2013. Geodetic imaging with airborne LiDAR: the Earth's surface revealed[J]. Reports on Progress in Physics, 76: 1-24
Goodbody T R H, Coops N C, Marshall P L, et al. 2017. Unmanned aerial systems for precision forest inventory purposes: a review and case study[J]. Forestry Chronicle, 93(1): 71-81.
Guo Q, Su Y, Hu T, et al. 2017. An integrated UAV-borne lidar system for 3D habitat mapping in three forest ecosystems across China[J]. International Journal of Remote Sensing, 38(8-10): 2954-2972.
Haala N, Hastedt H, Wolf K, et al. 2010. Digital photogrammetric camera evaluation - generation of digital elevation models[J]. Photogrammetrie Fernerkundung Geoinformation(2): 99-115.
Harpold A A, Guo Q, Molotch N, et al. 2014. LiDAR-derived snowpack data sets from mixed conifer forests across the Western United States[J]. Water Resources Research, 50(3): 2749-2755.
Harpold A A, Marshall J A, Lyon S W, et al. 2015. Laser vision: lidar as a transformative tool to advance critical zone science[J]. Hydrology and Earth System Sciences, 19(6): 2881-2897.
Hirschmuller H. 2005. Accurate and efficient stereo processing by semi-global matching and mutual information//Schmid C, Soatto S, Tomasi C. 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Vol 2, Proceedings[M]. San Diego, CA: IEEE, 807-814
Jarnstedt J, Pekkarinen A, Tuominen S, et al. 2012. Forest variable estimation using a high-resolution digital surface model[J]. Isprs Journal of Photogrammetry and Remote Sensing, 74: 78-84.
Jin X, Liu S, Baret F, et al. 2017. Estimates of plant density of wheat crops at emergence from very low altitude UAV imagery[J]. Remote Sensing of Environment, 198: 105-114.
Laurin G V, Chen Q, Lindsell J A, et al. 2014. Above ground biomass estimation in an African tropical forest with lidar and hyperspectral data[J]. Isprs Journal of Photogrammetry and Remote Sensing, 89: 49-58.
Leberl F, Irschara A, Pock T, et al. 2010. Point clouds: Lidar versus 3D vision[J]. Photogrammetric Engineering and Remote Sensing, 76(10): 1123-1134.
Neugirg F, Stark M, Kaiser A, et al. 2016. Erosion processes in calanchi in the upper Orcia Valley, Southern Tuscany, Italy based on multitemporal high-resolution terrestrial LiDAR and UAV surveys[J]. Geomorphology, 269: 8-22.
Ota T, Ogawa M, Shimizu K, et al. 2015. Aboveground biomass estimation using structure from motion approach with aerial photographs in a seasonal tropical forest[J]. Forests, 6(11): 3882-3898.
Penner M, Woods M, Pitt D G. 2015. A comparison of airborne laser scanning and image point cloud derived tree size class distribution models in Boreal Ontario[J]. Forests, 6(11): 4034-4054.
Piermattei L, Carturan L, Guarnieri A. 2015. Use of terrestrial photogrammetry based on structure-from-motion for mass balance estimation of a small glacier in the Italian alps[J]. Earth Surface Processes and Landforms, 40(13): 1791-1802.
Reutebuch S E, Mcgaughey R J, Andersen H E, et al. 2003. Accuracy of a high-resolution lidar terrain model under a conifer forest canopy[J]. Canadian Journal of Remote Sensing, 29(5): 527-535.
Santoso F, Garratt M A, Pickering M R, et al. 2016. 3D mapping for visualization of rigid structures: a review and comparative study[J]. Ieee Sensors Journal, 16(6): 1484-1507.
Shellberg J G, Spencer J, Brooks A P, et al. 2016. Degradation of the Mitchell River fluvial megafan by alluvial gully erosion increased by post-European land use change, Queensland, Australia[J]. Geomorphology, 266: 105-120.
Snavely N, Seitz S M, Szeliski R. 2008. Modeling the world from Internet photo collections[J]. International Journal of Computer Vision, 80(2): 189-210.
Stennett T A. 2004. Lidar: strap in tight, and prepare to go vertical[J]. Photogrammetric Engineering and Remote Sensing, 70(5): 545-548.
Stepper C, Straub C, Pretzsch H. 2015. Using semi-global matching point clouds to estimate growing stock at the plot and stand levels: application for a broadleaf-dominated forest in central Europe[J]. Canadian Journal of Forest Research, 45(1): 111-123.
St-Onge B, Audet F, Begin J. 2015. Characterizing the height structure and composition of a boreal forest using an individual tree crown approach applied to photogrammetric point clouds[J]. Forests, 6(11): 3899-3922.
St-Onge B, Vega C, Fournier R A, et al. 2008. Mapping canopy height using a combination of digital stereo-photogrammetry and lidar[J]. International Journal of Remote Sensing, 29(11): 3343-3364.
Tarolli P. 2014. High-resolution topography for understanding Earth surface processes: Opportunities and challenges[J]. Geomorphology, 216: 295-312.
Tinkham W T, Smith A M S, Hoffman C, et al. 2012. Investigating the influence of LiDAR ground surface errors on the utility of derived forest inventories[J]. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere, 42(3): 413-422.
Turner D, Lucieer A, Wallace L. 2014. Direct georeferencing of ultrahigh-resolution UAV imagery[J]. Ieee Transactions on Geoscience and Remote Sensing, 52(5): 2738-2745.
Turner D, Lucieer A, Watson C. 2012. An automated technique for generating georectified mosaics from ultra-high resolution unmanned aerial vehicle (UAV) imagery, based on structure from motion (SfM) point clouds[J]. Remote Sensing, 4(5): 1392-1410.
Wehr A, Lohr U. 1999. Airborne laser scanning - an introduction and overview[J]. Isprs Journal of Photogrammetry and Remote Sensing, 54(2-3): 68-82.
White J C, Coops N C, Wulder M A, et al. 2016. Remote sensing technologies for enhancing forest inventories: a review[J]. Canadian Journal of Remote Sensing, 42(5): 619-641.
White J C, Wulder M A, Vastaranta M, et al. 2013. The utility of image-based point clouds for forest inventory: a comparison with airborne laser scanning[J]. Forests, 4(3): 518-536.
White J, Stepper C, Tompalski P, et al. 2015. Comparing ALS and image-based point cloud metrics and modelled forest inventory attributes in a complex coastal forest environment[J]. Forests, 6(10): 3704.
狄颖辰, 陈云坪, 陈莹莹, 等. 2011. 无人机图像拼接算法综述[J]. 计算机应用, 31(1): 170-174. [Di Y C, Chen Y P, Chen Y Y, et al. 2011. Survey on image mosaic algorithm of unmanned aerial vehicle [J]. Journal of Computer Applications, 31(1): 170-174.]
郭庆华, 刘瑾, 李玉美, 等. 2016b. 生物多样性近地面遥感监测:应用现状与前景展望[J]. 生物多样性, 24(11): 1249-1266. [Guo Q H, Liu J, Li Y M, et al. 2016b. A near-surface remote sensing platform for biodiversity monitoring: perspectives and prospects [J]. Biodiversity Science, 24(11): 1249-1266.]
郭庆华, 刘瑾, 陶胜利, 等. 2014. 激光雷达在森林生态系统监测模拟中的应用现状与展望[J]. 科学通报, 59(6): 459-478. [Guo Q H, Liu J, Tao S L, et al. 2014. Perspectives and prospects of LiDAR in forest ecosystem monitoring and modeling [J]. Chin Sci Bull (Chin Ver), 59(6): 459-478.]
郭庆华, 吴芳芳, 胡天宇, 等. 2016a. 无人机在生物多样性遥感监测中的应用现状与展望[J]. 生物多样性, 24(11): 1267-1278. [Guo Q H, Wu F F, Hu T Y, et al. 2016a. Perspectives and prospects of unmanned aerial vehicle in remote sensing monitoring of biodiversity [J]. Biodiversity Science, 24(11): 1267-1278.]
李德仁, 刘良明, 胡晓沁. 2001. 1996-2000年中国摄影测量与遥感进展[J]. 测绘学报, 30(2): 117-126. [Li D R, Liu L M, Hu X Q. 2001. Development of photogrammetry and remote sensing in China from 1996 to 2000 [J]. Acta Geodaetica et Cartographica Sinica, 30(2): 117-126.]
李德仁. 2000. 摄影测量与遥感的现状及发展趋势[J]. 武汉测绘科技大学学报, 25(1): 1-6. [Li D R. 2000. Towards photogrammetry and remote sensing: status and future development [J]. Journal of Wuhan Technical University of Surveying and Mapping, 25(1): 1-6.]
刘清旺, 李世明, 李增元, 等. 2017. 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, 53(7): 134-148. [Liu Q W, Li S M, Li Z Y, et al. 2017. Review on the applications of UAV-based LiDAR and photogrammetry in forestry [J]. Scientia Silvae Sinicae, 53(7): 134-148.]
彭大雷, 许强, 董秀军, 等. 2017. 无人机低空摄影测量在黄土滑坡调查评估中的应用[J]. 地球科学进展,32(3): 319-330. [Peng D L, Xu Q, Dong X J, et al. 2017. Application of unmanned aerial vehicles low-altitude photogrammetry in investigation and evaluation of loess landslide [J]. Advances in Earth Science,32(3): 319-330.]
王帅永, 唐川, 何敬, 等. 2016. 无人机在强震区地质灾害精细调查中的应用研究[J]. 工程地质学报, 24(4): 713-719.[Wang S Y, Tang C, He J, et al. 2016. Use of unmanned aerial vehicle for precise investigation of geological hazard in strong seismic zone [J]. Journal of Engineering Geology, 24(4): 713-719.]
于景鑫, 刘金成, 冯仲科. 2015. 森林盗伐的数字摄影测量方法及应用[J]. 南京林业大学学报(自然科学版), 39(5): 87-92. [Yu J X, Liu J C, Feng Z K. 2015. Statistical methods and application of forest logging digital photogrammetry [J]. Journal of Nanjing Forestry University (Natural Sciences Edition), 39(5): 87-92.]
张盈松, 刘时银, 上官冬辉. 2012. 地面多基线数字摄影测量在冰川测量中的应用——以博格达峰黑沟8号冰川为例[J]. 冰川冻土, 34(5): 1184-1189. [Zhang Y S, Liu S Y, Shangguan D H. 2012. Ground multi-baseline digital photogrammetry for surveying the Heigou Glacier No. 8 in the Mt. Bogda, Tianshan Mountains [J]. Journal of Glaciology and Geocryology, 34(5): 1184-1189.]
张祖勋, 柯涛, 郭大海, 等. 2009. 数字摄影测量网格在汶川大地震中的快速响应[J]. 中国工程科学,11(6): 54-62. [Zhang Z X, Ke T, Guo D H, et al. 2009. The application of digital photogrammetry grid in the fast response for Wenchuan earthquake rescue [J]. Engineering Sciences, 11(6): 54-62.]
张祖勋, 吴媛. 2015. 摄影测量的信息化与智能化[J]. 测绘地理信息,40(4): 1-5. [Zhang Z X, Wu Y. 2015. Informatization and intellectualization of photogrammetry [J]. Journal of Geomatics, 40(4): 1-5.]
张祖勋, 张剑清, 张力. 2000. 数字摄影测量发展的机遇与挑战[J]. 武汉测绘科技大学学报, 25(1): 7-11. [Zhang Z X, Zhang J Q, Zhang L. 2000. Opportunities and challenges for development of digital photogrammetry [J]. Journal of Wuhan Technical University of surveying and mapping, 25(1): 7-11.]
周明, 邱凌云. 2018. 无人机技术在线路边坡病害巡查中的研究与应用[J]. 测绘地理信息, 43(4): 62-64.[Zhou M, Qiu L Y, 2018. Research and application of unmanned aerial vehicle technology in slope disease inspection by line [J]. Journal of Geomatics, 43(4), 62-64]
朱锋, 肖晖, 魏亚男. 2014. 无人机遥感影像镶嵌技术综述[J]. 计算机工程与应用, 50(15): 38-41. [Zhu F, Xiao H, Wei Y N. 2014. Summary of UAV remote sensing image mosaicking technology [J]. Computer Engineering and Applications, 50(15): 38-41.]
稿件与作者信息
罗达1, 2,4
LUO Da1, 2,4
林杭生1, 2,3
LIN Henry1, 2,3
林杭生,E-mail: henrylin@psu.edu
金钊1,2
JIN Zhao1,2
郑涵1,2
ZHENG Han1,2
宋怡1
SONG Yi1
冯立1,2,4
FENG Li1,2,4
郭庆华5
GUO Qinghua5
出版历史
出版时间: 2018年10月25日 (版本2
参考文献列表中查看
地球环境学报
Journal of Earth Environment