劉清旺 李世明 李增元 符利勇 胡凱龍,2
(1.中國林業(yè)科學(xué)研究院資源信息研究所 北京 100091; 2.中國礦業(yè)大學(xué)地球科學(xué)與測繪工程學(xué)院 北京 100083)
無人機(jī)激光雷達(dá)與攝影測量林業(yè)應(yīng)用研究進(jìn)展*
劉清旺1李世明1李增元1符利勇1胡凱龍1,2
(1.中國林業(yè)科學(xué)研究院資源信息研究所 北京 100091; 2.中國礦業(yè)大學(xué)地球科學(xué)與測繪工程學(xué)院 北京 100083)
森林空間結(jié)構(gòu)及動(dòng)態(tài)變化規(guī)律對(duì)森林經(jīng)營管理、生態(tài)環(huán)境建模等具有重要意義,無人機(jī)激光雷達(dá)與攝影測量能夠獲取豐富的森林空間結(jié)構(gòu)和類型信息,在單木、林分尺度森林環(huán)境長時(shí)間序列監(jiān)測方面具有無可比擬的優(yōu)勢。無人機(jī)激光雷達(dá)系統(tǒng)一般搭載多回波/全波形激光掃描儀,配備高精度全球?qū)Ш叫l(wèi)星系統(tǒng)&慣性測量單元 (GNSS & IMU)等傳感器,以保證激光脈沖回波信號(hào)的幾何定位精度。無人機(jī)攝影測量系統(tǒng)通常搭載可見光(RGB)/多光譜相機(jī),配備低精度GNSS & IMU,通過高重疊率航片的三維重建算法自動(dòng)解算航片內(nèi)外方位元素,生成具有相對(duì)參考坐標(biāo)的圖像及點(diǎn)云,采用地面控制點(diǎn)(GCPs)、參考影像等方式進(jìn)行幾何精校正,對(duì)于連續(xù)覆蓋的森林區(qū)域,使用高精度GNSS、穩(wěn)定平臺(tái)等可以提高圖像匹配精度。通過單木分割法可以提取單木結(jié)構(gòu)信息,從激光雷達(dá)點(diǎn)云或攝影測量重建點(diǎn)云中識(shí)別樹冠頂點(diǎn)、樹冠邊界、位置等屬性,也可以將點(diǎn)云投影到體元空間或者生成冠層高度模型(CHM),在此基礎(chǔ)上識(shí)別單木特征。林分結(jié)構(gòu)信息提取常采用高度分布法,從點(diǎn)云中直接計(jì)算高度分位數(shù)、回波指數(shù)等點(diǎn)云特征量,或者按照指定的高度間隔生成頻率或強(qiáng)度合成波形,計(jì)算波形分位數(shù)、波形前沿、波形后沿等波形特征量,根據(jù)點(diǎn)云特征量、波形特征量與地面測量值之間的關(guān)系估測森林結(jié)構(gòu)參數(shù)。激光雷達(dá)點(diǎn)云和攝影測量重建點(diǎn)云均能用于提取林下地形,對(duì)于低郁閉度區(qū)域二者相差不大,對(duì)于高郁閉度區(qū)域攝影測量重建點(diǎn)云提取的林下地形精度較低。多時(shí)相無人機(jī)激光雷達(dá)和攝影測量相結(jié)合,可以監(jiān)測人工修枝、擇伐、火災(zāi)、病蟲害等引起的森林結(jié)構(gòu)變化以及枝葉生長、落葉等物候變化。無人機(jī)激光雷達(dá)與攝影測量提取的森林結(jié)構(gòu)參數(shù)精度受采集方式、數(shù)據(jù)處理算法、森林生長季節(jié)、地形等因素影響,尚未形成適合林業(yè)推廣應(yīng)用的成熟技術(shù)體系。無人機(jī)系統(tǒng)飛行應(yīng)當(dāng)遵照國家/當(dāng)?shù)胤煞ㄒ?guī)以及相關(guān)規(guī)定條款的約束,我國按照空機(jī)質(zhì)量、起飛全重等指標(biāo)對(duì)無人機(jī)進(jìn)行分類管理。未來無人機(jī)數(shù)據(jù)獲取與處理系統(tǒng)將更加智能化、微型化、低成本化,更好地滿足林業(yè)應(yīng)用業(yè)務(wù)需求。
無人機(jī); 激光雷達(dá); 攝影測量; 點(diǎn)云; 森林
森林是陸地生態(tài)系統(tǒng)的主體,在維護(hù)區(qū)域生態(tài)環(huán)境及全球碳平衡、緩解全球氣候變化等方面發(fā)揮著不可替代的作用 (Hyyppaetal., 2008),森林空間結(jié)構(gòu)及動(dòng)態(tài)變化規(guī)律研究對(duì)森林經(jīng)營管理、生態(tài)環(huán)境建模及碳循環(huán)分析具有重要意義 (Waseretal., 2015; Dandoisetal., 2015)。森林空間結(jié)構(gòu)的自然變化特征主要受樹木生長狀況控制(如樹高的增加、樹冠尺寸的增加、樹冠之間空隙的減少等),年積溫、水分脅迫、立地條件等外部因素會(huì)直接影響樹木的生長狀況。另外,人工修枝、間伐/擇伐、森林病蟲害傳播、外來物種入侵、林火、風(fēng)災(zāi)、冰凍雪災(zāi)等均會(huì)直接影響森林結(jié)構(gòu)變化 (McElhinnyetal., 2005)。
傳統(tǒng)森林空間結(jié)構(gòu)地面測定方法僅能得到一些點(diǎn)上數(shù)據(jù),很難及時(shí)獲取區(qū)域范圍森林參數(shù)的空間分布信息,而快速發(fā)展的遙感技術(shù)能夠滿足此應(yīng)用需求 (Maseketal., 2015; Waseretal., 2015; Ginzleretal., 2015)。遙感圖像光譜信息具有良好的綜合性和現(xiàn)勢性,能夠反映森林空間分布的變化特征,激光雷達(dá) (light detection and ranging, LiDAR) 在森林垂直結(jié)構(gòu)測量方面具有無可比擬的優(yōu)勢,將遙感圖像、激光雷達(dá)數(shù)據(jù)與地面測量得到的樣點(diǎn)數(shù)據(jù)相結(jié)合,可以更精準(zhǔn)地獲取大范圍森林類型及三維空間結(jié)構(gòu)的分布特征,通過多時(shí)相遙感數(shù)據(jù)序列分析森林結(jié)構(gòu)的動(dòng)態(tài)變化規(guī)律。
近年來,無人機(jī) (unmanned aerial vehicle, UAV; unmanned aerial system, UAS) 遙感技術(shù)的快速發(fā)展,在全球范圍內(nèi)引起了廣泛關(guān)注,很多組織機(jī)構(gòu)或個(gè)人開始研制、集成多種形式的無人機(jī)遙感系統(tǒng) (Floreanoetal., 2015; 范承嘯等, 2009; 金偉等, 2009; 李德仁等, 2014),用于快速地形測量、地表變化監(jiān)測等業(yè)務(wù) (Schereretal., 2008; Xiangetal., 2011; Zhouetal., 2012; Wattsetal., 2012; Colominaetal., 2014)。針對(duì)我國森林資源調(diào)查業(yè)務(wù)來說,需要在全國范圍內(nèi)定期觀測森林樣地,無人機(jī)遙感具有廣闊的應(yīng)用前景,特別是對(duì)于不易到達(dá)的區(qū)域,無人機(jī)具有更為顯著的優(yōu)勢。
無人機(jī)遙感系統(tǒng)主要由無人飛行器、遙感載荷、地面控制站和通信數(shù)據(jù)鏈組成 (Wattsetal., 2012; Colominaetal., 2014; Pajares, 2015),最早用于輻射污染、危險(xiǎn)區(qū)、惡劣環(huán)境探測任務(wù) (Wattsetal., 2012),后來用于監(jiān)測森林 (Hernández-Clementeetal., 2012; 許子乾等, 2015)、農(nóng)作物 (Zarco-Tejadaetal., 2013; Córcolesetal., 2013; 王利民等, 2013; 汪沛等, 2014; 張波等, 2015)、土地變化 (Mesas-Carrascosaetal., 2014)、野生動(dòng)物 (Ditmeretal., 2015; Liuetal., 2015)、冰川 (Immerzeeletal., 2014)、溫室氣體 (Bermanetal., 2012)、災(zāi)害 (Martínez-de Diosetal., 2006; Olleroetal., 2006; Karmaetal., 2015; 周潔萍等, 2008; 胡根生等, 2014; 張?jiān)龅? 2015)、環(huán)境保護(hù) (楊海軍等, 2015; 洪運(yùn)富等, 2015) 等。無人機(jī)與有人機(jī)、衛(wèi)星遙感應(yīng)用相比,使用成本低,空間分辨率更高,在選擇適用載荷、時(shí)間及空間分辨率方面更加靈活 (Dunfordetal., 2009; Zhangetal., 2012; 秦博等, 2002)。
1.1 無人機(jī)類型
無人機(jī)一般分為多旋翼、固定翼和無人直升機(jī)等類型 (王峰等, 2010; 李德仁等, 2014)。多旋翼無人機(jī)能夠垂直起降,負(fù)載能力有限,旋翼數(shù)量從3個(gè)到十幾個(gè)不等,隨著旋翼數(shù)量增多,無人機(jī)的飛行和起降穩(wěn)定性增加,但是也增加了制造成本; 固定翼無人機(jī)的起飛方式包括手拋式、彈射式、跑道式等,其體積、巡航速度、負(fù)載能力差異很大,固定翼轉(zhuǎn)彎時(shí)不如多旋翼靈活,不適合狹窄空間的觀測任務(wù); 無人直升機(jī)同樣能夠垂直起降,負(fù)載能力通常優(yōu)于多旋翼無人機(jī),但是可操控性相對(duì)困難些 (Colominaetal., 2014)。Watts等(2012)按照無人機(jī)尺寸、航時(shí)和負(fù)載等特征,將無人機(jī)分為微型、垂直起降型、低空短航時(shí)、低空長航時(shí)、中空長航時(shí)、高空長航時(shí)等類型,搭載不同類型的傳感器執(zhí)行遙感任務(wù)。
微/輕型無人機(jī)一般指質(zhì)量/體積小的無人機(jī),優(yōu)點(diǎn)是攜帶方便,能夠高效獲取精細(xì)尺度的高空間和高時(shí)間分辨率數(shù)據(jù),例如傳統(tǒng)森林樣地面積或幾平方千米面積,用于監(jiān)測森林冠層間隙分布、樹冠尺寸、森林病害程度等 (Lehmannetal., 2015; Getzinetal., 2014; Díaz-Varelaetal., 2015),能夠測量三維結(jié)構(gòu)的無人機(jī)越來越多地用于森林結(jié)構(gòu)的變化分析 (Dandoisetal., 2013; Liseinetal., 2013; Wallaceetal., 2014a; 2014b; 2014c; 2016)。
無人機(jī)飛行應(yīng)當(dāng)遵照國家/當(dāng)?shù)胤煞ㄒ?guī)以及相關(guān)規(guī)定條款的約束。我國出臺(tái)了一系列無人機(jī)系統(tǒng)運(yùn)行及駕駛員管理相關(guān)試行/暫行規(guī)定,按照空機(jī)質(zhì)量、起飛全重等指標(biāo)對(duì)無人機(jī)進(jìn)行分類管理,其中對(duì)于Ⅰ類 (起飛全重≤1.5 kg) 和Ⅱ類 (起飛全重≤7 kg) 無人機(jī),除重點(diǎn)區(qū)域和機(jī)場凈空區(qū)外,在視距內(nèi) (半徑500 m、相對(duì)高度低于120 m的區(qū)域) 運(yùn)行時(shí)無須證照管理,其他類無人機(jī)應(yīng)當(dāng)使用電子圍欄并接入無人機(jī)云。
1.2 無人機(jī)遙感載荷
無人機(jī)遙感載荷類似于有人機(jī),包括可見光相機(jī)、多光譜相機(jī)、高光譜成像儀、熱紅外成像儀、激光雷達(dá)、微波雷達(dá)等,但是要求體積更小、自動(dòng)化程度更高。無人機(jī)可見光載荷應(yīng)用廣泛 (Dandoisetal., 2010; 2013; Grenzd?rfferetal., 2012),從普通相機(jī)到專業(yè)航空相機(jī)均可搭載到無人機(jī)平臺(tái)上,特別是計(jì)算機(jī)視覺圖像處理技術(shù)的快速發(fā)展,大大提升了光學(xué)圖像數(shù)據(jù)的處理能力 (呂書強(qiáng)等, 2007)。無人機(jī)多光譜載荷提供了可見光、近紅外等光譜波段,能夠反映植被健康的光譜信息 (Bendigetal., 2012; Kelceyetal., 2012)。高光譜比多光譜具有更高的光譜分辨率,用于提取植被生化組分信息 (Rufinoetal., 2005; Zarco-Tejadaetal., 2013)。熱紅外載荷對(duì)溫度變化反應(yīng)敏感,用于林火監(jiān)測或軍事偵查等目的 (Rufinoetal., 2005; Scholtzetal., 2011)。無人機(jī)已經(jīng)能夠搭載微型激光雷達(dá),用于直接測量植被三維結(jié)構(gòu) (Nagaietal., 2004; Choietal., 2009)。通過微波雷達(dá)P波段和X波段組合能夠獲取森林結(jié)構(gòu)信息,但是需要解決雷達(dá)微型化問題,才能用于無人機(jī)遙感 (Essenetal., 2012; Remyetal., 2012)。
1.3 無人機(jī)激光雷達(dá)系統(tǒng)
無人機(jī)激光雷達(dá)系統(tǒng)通常配備激光掃描儀、高精度全球?qū)Ш叫l(wèi)星系統(tǒng)&慣性測量單元 (global navigation satellite system, GNSS & inertial measurement unit, IMU) 和光學(xué)相機(jī)等載荷,激光掃描儀能夠獲取地物的三維點(diǎn)云數(shù)據(jù),高精度GNSS & IMU提供姿態(tài)和位置信息,用于解算激光點(diǎn)三維位置,光學(xué)相機(jī)用于地物類型識(shí)別以及結(jié)構(gòu)信息提取。隨著微型激光掃描儀、GNSS & IMU集成系統(tǒng)的快速發(fā)展,出現(xiàn)了多種輕小型無人機(jī)激光雷達(dá)系統(tǒng) (Nagaietal., 2004; 2009; Jaakkolaetal., 2010; Linetal., 2011),采用高精度GNSS & IMU是提高激光雷達(dá)定位精度的關(guān)鍵,集成GNSS & IMU的混合測量單元 (hybrid measurement unit, HMU) 已經(jīng)能夠達(dá)到厘米級(jí)定位精度 (Rehaketal., 2013),另外,低成本GNSS & IMU系統(tǒng)精度較低,采用GNSS雙天線可以有效改善航向精度。無人機(jī)自動(dòng)巡航系統(tǒng)能夠?qū)崟r(shí)地獲取飛行器位置、速度和姿態(tài)信息,并反饋給飛行控制系統(tǒng),使得飛行器沿著預(yù)定航線飛行,自動(dòng)巡航開源框架包括Paparazzi、OpenPilot、ArduPilot等 (Colominaetal., 2014)。
幾種典型無人機(jī)激光雷達(dá)的主要技術(shù)指標(biāo)見表1,另外,也出現(xiàn)了一些其他形式的無人機(jī)激光雷達(dá)系統(tǒng),由于缺少詳細(xì)技術(shù)資料未列出。無人機(jī)激光雷達(dá)系統(tǒng)通常搭載數(shù)碼相機(jī),同時(shí)獲取激光雷達(dá)和光學(xué)圖像數(shù)據(jù),提供精細(xì)的結(jié)構(gòu)信息和光譜信息 (Nagaietal., 2004; 2009; Jaakkolaetal., 2010; Linetal., 2011)。Nagai等(2004; 2009) 開發(fā)了一種無人機(jī)激光雷達(dá)系統(tǒng),包括GNSS & IMU系統(tǒng)、低頻激光掃描儀、單反相機(jī)等,用于提取DSM和紋理特征。Jaakkola等(2010) 和Lin等(2011) 設(shè)計(jì)了一種輕小型無人機(jī)激光雷達(dá)系統(tǒng) (FGI Sensei),包括GNSS & IMU集成系統(tǒng)、激光掃描儀、CCD相機(jī)、光譜儀和熱紅外相機(jī),采用模塊化設(shè)計(jì),可以任意組合不同遙感傳感器,并能用于車載移動(dòng)制圖。Wallace等 (2012; 2014a; 2014b; 2014c; 2016) 研制了一種低成本無人機(jī)激光雷達(dá)系統(tǒng) (TerraLuma),包括GNSS & IMU系統(tǒng)、低頻激光掃描儀、高清錄像機(jī)等,用于森林資源清查。Gottfried等(2015) 和 Amon等(2015)介紹了一種RIEGL的無人機(jī)激光雷達(dá)系統(tǒng)(RIEGL VUX-SYS),包括GNSS & IMU集成系統(tǒng)、高頻激光掃描儀、可見光相機(jī)等,面向林業(yè)等多種行業(yè)應(yīng)用。另外,Velodyne 推出的激光雷達(dá)可以用于車載平臺(tái) (Glennieetal., 2013; Rejasetal., 2015) 和無人機(jī)平臺(tái) (Colominaetal., 2014; Tulldahletal., 2014; 2015),引起了廣泛關(guān)注。
表1 典型無人機(jī)激光雷達(dá)的主要技術(shù)指標(biāo)
無人機(jī)激光雷達(dá)與有人機(jī)相比,可以獲取很高采樣密度的觀測數(shù)據(jù),主要為離散的單回波或多回波點(diǎn)云;有些無人機(jī)激光雷達(dá)能夠獲取全波形數(shù)據(jù),但在應(yīng)用中多采用在線波形處理方式得到多回波點(diǎn)云數(shù)據(jù) (Gottfriedetal., 2015; Amonetal., 2015)。
1.4 無人機(jī)攝影測量系統(tǒng)
無人機(jī)攝影測量系統(tǒng)通常配備普通相機(jī)、低精度GNSS & IMU等載荷 (孫杰等, 2003; 崔紅霞等, 2005; 賈建軍等, 2006; 李宇昊, 2007),普通相機(jī)一般具備自動(dòng)連拍功能,低精度GNSS & IMU多用于自動(dòng)巡航,不適于圖像精確定位。無人機(jī)攝影測量平臺(tái)可以選擇固定翼、多旋翼或直升機(jī),固定翼飛行速度快,適合于較大區(qū)域測量任務(wù); 多旋翼或直升機(jī)飛行速度慢,轉(zhuǎn)彎半徑小,適合于小區(qū)域精細(xì)測量任務(wù) (Shahbazietal., 2015)。Breckenridge等(2011; 2012) 比較了固定翼和直升機(jī)2種無人機(jī),發(fā)現(xiàn)不同無人機(jī)平臺(tái)得到的植被覆蓋度估測結(jié)果與地面測量值的一致性很好。
無人機(jī)攝影測量一般通過高重疊率圖像的自動(dòng)匹配處理算法,自動(dòng)解算圖像內(nèi)外方位元素,生成具有相對(duì)參考坐標(biāo)的點(diǎn)云,幾何精校正可采用地面控制點(diǎn)、參考影像&地形、航跡曲線匹配等方法 (Dandoisetal., 2010; 2013; Zhangetal., 2016)?;诘孛婵刂泣c(diǎn)的方法通過高精度GPS或全站儀等儀器對(duì)地物特征點(diǎn)進(jìn)行測量,獲取控制點(diǎn)三維坐標(biāo); 基于參考影像&地形的方法通過參考影像獲取典型特征對(duì)象的平面坐標(biāo),從地形數(shù)據(jù)中獲取相應(yīng)特征對(duì)象的高程坐標(biāo); 航跡曲線匹配方法使用無人機(jī)航跡與重建圖像的相機(jī)位置軌跡匹配,由于圖像重建點(diǎn)云的幾何特征無法與參考數(shù)據(jù)幾何特征進(jìn)行精確匹配,因此校正精度一般為亞米級(jí)或米級(jí) (Dandoisetal., 2010; 2013)。如Dandois等 (2010) 通過參考正射影像和激光雷達(dá)DEM 對(duì)圖像重建點(diǎn)云進(jìn)行幾何校正 (平面定位精度<1.5 m,高程定位精度0.6~0.9 m); Lisein等 (2013) 通過參考正射影像和激光雷達(dá)DSM 對(duì)圖像重建點(diǎn)云進(jìn)行幾何精校正; Zhang等(2016)通過地面GNSS測量的控制點(diǎn)對(duì)圖像重建點(diǎn)云進(jìn)行幾何校正 (定位精度0.32~0.69 m); Dandois等(2013) 通過地面控制點(diǎn)和航跡曲線匹配方式對(duì)圖像重建點(diǎn)云進(jìn)行幾何校正 (定位精度0.4~1.4 m)。
對(duì)于連續(xù)覆蓋的森林區(qū)域,當(dāng)從圖像中很難識(shí)別典型同名點(diǎn)時(shí),通過高精度GNSS & IMU數(shù)據(jù)可以有效提高匹配效果 (Turneretal., 2014; Zarco-Tejadaetal., 2014; Wallaceetal., 2016),在缺少高精度GNSS & IMU情況下,通過穩(wěn)定平臺(tái)可以提高定位精度 (洪宇等, 2008),或者通過圖像和激光雷達(dá)數(shù)據(jù)自動(dòng)配準(zhǔn)方法提高定位精度 (Yangetal., 2015)。例如Zarco-Tejada等(2014) 使用圖像同步GPS位置來保證圖像三維重建點(diǎn)云定位精度; Wallace等(2016) 使用高精度GNSS數(shù)據(jù)得到較高的圖像定位精度,不需要使用參考數(shù)據(jù)進(jìn)行幾何校正。另外,低精度GNSS在圖像三維重建中可輔助篩選初始匹配圖像,提高處理效率,但是無法保證定位精度(Liseinetal., 2013)。
無人機(jī)獲取森林三維空間結(jié)構(gòu)的常用方式有2種 (Wallaceetal., 2016): 一是激光掃描 (Jaakkolaetal., 2010; Wallaceetal., 2012); 二是攝影測量 (Dandoisetal., 2013; Liseinetal., 2013)。激光掃描系統(tǒng)通過發(fā)射激光脈沖并接收探測目標(biāo)后向散射信號(hào)的方式來測量地物的三維空間結(jié)構(gòu),攝影測量通過計(jì)算機(jī)視覺等圖像三維重建技術(shù)得到點(diǎn)云,激光雷達(dá)和攝影測量點(diǎn)云能夠用于提取典型的森林結(jié)構(gòu)信息 (Jaakkolaetal., 2010; Wallaceetal., 2014a; 2014b; 2014c; 2016)。
2.1 無人機(jī)激光雷達(dá)森林結(jié)構(gòu)
激光雷達(dá)能夠穿透森林冠層,直接測量森林三維結(jié)構(gòu),在森林經(jīng)營管理與生態(tài)系統(tǒng)研究中具有廣闊應(yīng)用前景 (Hyyppaetal., 2008)。傳統(tǒng)機(jī)載激光雷達(dá)數(shù)據(jù)獲取成本高,獲取時(shí)間和范圍受空域政策限制,同時(shí)氣候條件進(jìn)一步限制了有效觀測時(shí)間,很難獲取多時(shí)相激光雷達(dá)數(shù)據(jù),對(duì)于森林災(zāi)害、落葉監(jiān)測等時(shí)效性強(qiáng)的測量任務(wù)來說無法滿足需求。無人機(jī)激光雷達(dá)操作靈活,特別是輕小型無人機(jī)具有便攜性,可為森林結(jié)構(gòu)研究和森林資源調(diào)查提供一個(gè)有效的技術(shù)手段。
無人機(jī)激光雷達(dá)數(shù)據(jù)獲取與預(yù)處理流程包括航線設(shè)計(jì)、數(shù)據(jù)采集、GNSS & IMU數(shù)據(jù)預(yù)處理、點(diǎn)云數(shù)據(jù)解算、點(diǎn)云分類等 (Wallaceetal., 2012)。航線設(shè)計(jì)需要考慮點(diǎn)云密度、飛行高度、航線間隔、激光掃描角、激光最大測距范圍、地形等因素,按照飛行區(qū)域,設(shè)定飛行器進(jìn)入點(diǎn)和飛出點(diǎn),確定航線方向。數(shù)據(jù)采集時(shí)選擇控制點(diǎn)架設(shè)GNSS參考站,實(shí)時(shí)監(jiān)測無人機(jī)工作情況,根據(jù)無人機(jī)續(xù)航能力及時(shí)調(diào)整返航時(shí)間,注意風(fēng)向和風(fēng)速的變化等外部環(huán)境因素。GNSS & IMU數(shù)據(jù)預(yù)處理根據(jù)GNSS流動(dòng)站和參考站數(shù)據(jù)進(jìn)行差分解算,并將GNSS精確位置數(shù)據(jù)與IMU姿態(tài)數(shù)據(jù)合成。點(diǎn)云數(shù)據(jù)解算根據(jù)激光掃描數(shù)據(jù)、精確位置和姿態(tài)數(shù)據(jù)得到激光點(diǎn)的精確位置。點(diǎn)云分類是將點(diǎn)云分為地面點(diǎn)、植被點(diǎn)等不同類別,用于森林結(jié)構(gòu)信息提取。
無人機(jī)與有人機(jī)激光雷達(dá)提取森林結(jié)構(gòu)信息的方法類似,激光雷達(dá)波形數(shù)據(jù)通過波形分解可以得到點(diǎn)云數(shù)據(jù) (Liuetal., 2011),激光雷達(dá)點(diǎn)云數(shù)據(jù)提取森林結(jié)構(gòu)信息的方法可分為單木分割法和高度分布法2種(Vauhkonenetal., 2011; Kaartinenetal., 2012),單木分割法根據(jù)單木空間形狀特征識(shí)別樹冠頂點(diǎn)、樹冠邊界、位置等屬性 (Brandtbergetal., 2003; Maltamoetal., 2004; Kochetal., 2006; Lietal., 2012; 劉清旺等, 2008; 2010); 高度分布法 (也稱基于面積的方法) 根據(jù)冠層高度分布特征與地面測量值之間的關(guān)系估測森林結(jié)構(gòu)參數(shù),如生物量、蓄積量、胸高斷面積等 (Nsset, 2002; Holmgren, 2004; Packalénetal., 2007; Hudaketal., 2008)。單木分割法用于提取單木尺度結(jié)構(gòu)信息,進(jìn)而提取林分尺度結(jié)構(gòu)信息; 高度分布法多用于提取林分尺度結(jié)構(gòu)信息。
單木分割法的數(shù)據(jù)特征空間可以分為點(diǎn)云、體元和CHM 3種(Wallaceetal., 2014a),基于點(diǎn)云特征空間的方法直接使用原始/歸一化點(diǎn)云進(jìn)行分割,包括三角剖分算法、貝葉斯算法、局部最大值聚類算法等 (Alexander, 2009; Reitbergeretal., 2009; L?hivaaraetal., 2012; Lietal., 2012); 基于體元特征空間的方法將點(diǎn)云投影到體元空間,體元屬性為體元內(nèi)所包含的點(diǎn)個(gè)數(shù),采用形態(tài)學(xué)算法識(shí)別樹冠 (Wangetal., 2008; Vaughnetal., 2012); 基于CHM空間特征的方法使用CHM進(jìn)行分割,為了消除偽樹冠頂點(diǎn)和樹冠凹陷點(diǎn),分割之前一般進(jìn)行CHM平滑處理 (Brandtbergetal., 2003; 2007; Kochetal., 2006; Popescu, 2007; Yuetal., 2011; 劉清旺等, 2008; 2010)。另外,基于混合空間特征的方法使用多種空間特征進(jìn)行綜合分析,如在平滑后CHM識(shí)別局部最大值,將其用作點(diǎn)云特征空間K均值聚類的種子點(diǎn)進(jìn)行單木分割 (Morsdorfetal., 2003; Guptaetal., 2010)。
在無人機(jī)激光雷達(dá)提取單木尺度結(jié)構(gòu)信息方面,Jaakkola等(2010) 通過無人直升機(jī)激光雷達(dá)(FGI Sensei) 獲取高密度點(diǎn)云數(shù)據(jù) (點(diǎn)云密度為100~1 500點(diǎn)·m-2),采用基于CHM的單木分割法提取樹高和樹冠形狀,樹高估測精度很高 (標(biāo)準(zhǔn)差約30 cm),并由樹冠邊界內(nèi)點(diǎn)云數(shù)據(jù)生成26個(gè)特征量,通過隨機(jī)森林算法估測胸徑 (RMSE=2.1 cm); Lin等(2011)針對(duì)高密度點(diǎn)云數(shù)據(jù)提出了多尺度柵格化算法,用于識(shí)別單木樹冠特征; Wallace等(2014a) 采用多旋翼無人機(jī)激光雷達(dá)獲取了澳大利亞東南部4年生桉樹(Eucalyptus)林點(diǎn)云數(shù)據(jù) (50點(diǎn)·m-2),激光入射角限制為偏離天底方向±30°,通過分析5種單木分割法,發(fā)現(xiàn)基于CHM和原始點(diǎn)云的方法最優(yōu),能夠探測98%以上的樹木; Wallace等(2014b) 在桉樹林重復(fù)性飛行試驗(yàn)中發(fā)現(xiàn)單木位置 (平均偏差 < 0.48 m) 和樹高 (平均偏差 < 0.35 m) 精度很高,樹冠面積和樹冠體積依賴于分割算法。
高度分布法的數(shù)據(jù)特征空間可以分為點(diǎn)云和波形2種,基于點(diǎn)云特征空間的方法直接使用點(diǎn)云進(jìn)行分析,計(jì)算相關(guān)統(tǒng)計(jì)或指數(shù)特征量,如不同的高度分位數(shù)、平均值、標(biāo)準(zhǔn)差、峰度、偏度、首回波郁閉指數(shù)、冠形郁閉指數(shù)等 (Goodwinetal., 2006; Wallaceetal., 2014b); 基于波形特征空間的方法是在一定空間范圍內(nèi),按照指定的高度間隔進(jìn)行點(diǎn)云數(shù)據(jù)進(jìn)行統(tǒng)計(jì),生成頻率或強(qiáng)度合成波形,計(jì)算波形特征量,如波形分位數(shù)、波形峰值、波形前沿、波形后沿等。根據(jù)高度分布法提取的系列特征量,采用回歸分析、機(jī)器學(xué)習(xí)等算法間接估測相關(guān)森林參數(shù) (Zhaoetal., 2011; Gleasonaetal., 2012; García-Gutiérrezetal., 2015)。
在無人機(jī)激光雷達(dá)提取林分尺度結(jié)構(gòu)信息方面,Wallace等(2014b) 通過重復(fù)性飛行試驗(yàn)分析了基于點(diǎn)云的林分變量穩(wěn)定性,發(fā)現(xiàn)不同飛行方式的點(diǎn)云垂直分布差異很小,首回波的高度最大值很穩(wěn)定,高度平均值、標(biāo)準(zhǔn)層、峰度和偏度的偏差較小 (< 0.05 m),較低高度分位數(shù) (20%、 30%、 40%) 的偏差較大 (< 0.09 m); 首回波郁閉指數(shù)的偏差小于冠形郁閉指數(shù)的偏差。
無人機(jī)激光雷達(dá)飛行高度一般較低 (距離樹冠頂部10~50 m),可以獲取很高密度的點(diǎn)云數(shù)據(jù),通過點(diǎn)云抽稀算法可得到不同采樣密度的點(diǎn)云數(shù)據(jù),用于分析點(diǎn)云密度對(duì)樹冠探測結(jié)果的影響。Lin 等(2011) 通過對(duì)比分析原始點(diǎn)云 (100~1 500點(diǎn)·m-2) 和抽稀點(diǎn)云 (4點(diǎn)·m-2) 提取的樹高,發(fā)現(xiàn)高點(diǎn)云密度能夠明顯改善樹高被低估現(xiàn)象; Wallace等(2012) 采用多旋翼無人機(jī)激光雷達(dá)獲取了點(diǎn)云數(shù)據(jù) (50點(diǎn)·m-2),提取了單木位置、樹高、冠幅和樹冠面積,與抽稀點(diǎn)云 (8點(diǎn)·m-2) 提取的單木參量進(jìn)行對(duì)比分析,發(fā)現(xiàn)隨著點(diǎn)云密度增加,樹高和樹冠位置的標(biāo)準(zhǔn)差更小,探測到樹冠頂點(diǎn)的概率更大,冠幅容易受激光發(fā)散角和激光入射角的影響; Wallace等(2014a) 在桉樹林飛行試驗(yàn)中發(fā)現(xiàn)隨著點(diǎn)云密度 (從5點(diǎn)·m-2到50點(diǎn)·m-2) 增加,可以明顯減少單木漏檢率。
2.2 無人機(jī)攝影測量森林結(jié)構(gòu)
攝影測量通過圖像三維重建獲取植被表面的高度信息,由于攝影測量無法穿透森林冠層,很難獲得林下地形信息。圖像三維重建時(shí)要求圖像具有很高的重疊率、便于識(shí)別的同名點(diǎn)以及適中的圖像尺寸和空間分辨率 (狄穎辰等, 2011; 朱鋒等, 2014)。隨著低成本的輕小型無人機(jī)、高速連拍的數(shù)碼相機(jī)以及基于計(jì)算機(jī)視覺的三維重建算法的快速發(fā)展,無人機(jī)攝影測量成為一種具有更低成本的森林結(jié)構(gòu)測量方式,并同時(shí)提供不同森林類型的光譜信息,特別適合于獲取小區(qū)域內(nèi)的多時(shí)相觀測數(shù)據(jù) (Dandoisetal., 2013)。
常用的圖像三維重建方法主要包括運(yùn)動(dòng)結(jié)構(gòu)重建 (structure from motion, SfM) 和半全局匹配 (semi-global matching, SGM) 算法等,SfM將攝影測量和計(jì)算機(jī)視覺算法相結(jié)合,對(duì)于不同視角的重疊圖像,自動(dòng)提取圖像特征、匹配特征和光束平差,不需要GNSS & IMU數(shù)據(jù),一些常用圖像三維重建軟件包中封裝了SfM算法,適合于處理具有高重疊率的無人機(jī)圖像數(shù)據(jù) (Snavelyetal., 2008; Dandoisetal., 2010; Turneretal., 2012; Otaetal., 2015; St-Ongeetal., 2015); SGM使用代價(jià)函數(shù)約束像元匹配的概率,采用全方向路徑優(yōu)化提高匹配效率 (Hirshmuller, 2008; Gehrkeetal., 2011; 2012; Whiteetal., 2015; Penneretal., 2015),遙感軟件包 (remote sensing software package graz, RSG) 封裝了SGM算法。典型圖像三維重建軟件見表2。
表2 典型圖像三維重建軟件
圖像重建點(diǎn)云數(shù)據(jù)與激光雷達(dá)點(diǎn)云提取森林結(jié)構(gòu)信息的方法類似,包括單木尺度和林分尺度結(jié)構(gòu)信息提取相關(guān)方法。在無人機(jī)攝影測量提取單木尺度結(jié)構(gòu)信息方面,Dandois等(2010) 使用風(fēng)箏平臺(tái)搭載的相機(jī)獲取異齡林和同齡林航空圖像,通過Ecosynth重建三維點(diǎn)云,圖像重建CHM能夠用于估測樹高 (R2> 0.64),但是不如激光雷達(dá)CHM估測樹高的精度高 (R2> 0.82); Zarco-Tejada等(2014) 分析了固定翼無人機(jī)獲取的橄欖樹(Oleaeuropaea)近紅外圖像,通過Pix4UAV進(jìn)行三維重建,從圖像重建DSM中獲取樹木高度信息,與地面實(shí)測樹高相關(guān)性很好 (R2=0.83, RMSE=35 cm)。
在無人機(jī)攝影測量提取林分尺度結(jié)構(gòu)信息方面,Zahawi等(2015) 通過多旋翼無人機(jī)獲取了熱帶林航空圖像,采用Ecosynth生成圖像三維點(diǎn)云,根據(jù)CHM特征量 (高度均值、最小值、最大值、標(biāo)準(zhǔn)差、分位數(shù)等) 分析冠層高度、地上生物量等因子,并分析冠層孔隙度、粗糙度等因子,評(píng)價(jià)森林恢復(fù)情況并預(yù)測食果性鳥類的分布; Zhang等(2016) 通過多旋翼無人機(jī)獲取了亞熱帶常綠闊葉林航空圖像,采用Pix4dmapper生成圖像三維點(diǎn)云,根據(jù)不同尺度的CHM特征量 (高度均值、標(biāo)準(zhǔn)差、偏度、垂直分布比率、郁閉度等),分析物種豐富度、香農(nóng)多樣性指數(shù)、物種均勻度、林分?jǐn)嗝娣e等因子,發(fā)現(xiàn)無人機(jī)冠層變量能夠很好反映局地生物多樣性模式,林分?jǐn)嗝娣e與郁閉度呈正相關(guān); Ni等(2015)通過多旋翼無人機(jī)獲取了北方森林航空圖像,采用Agisoft Photoscan重建圖像三維點(diǎn)云,對(duì)比分析攝影測量CHM和激光雷達(dá)CHM,發(fā)現(xiàn)樣地尺度森林高度相關(guān)性很好 (R2=0.87, RMSE=1.9 m)。另外,White等(2015) 采用RSG對(duì)高空航空圖像進(jìn)行三維重建 (點(diǎn)云密度為12.27點(diǎn)·m-2),由于旁向重疊率太低,僅使用航向圖像進(jìn)行匹配,圖像重建點(diǎn)云減去激光雷達(dá)DEM得到地形歸一化點(diǎn)云,按照坡度和郁閉度變化進(jìn)行分層分析,圖像重建點(diǎn)云與激光雷達(dá)點(diǎn)云特征量在統(tǒng)計(jì)上差異顯著,胸徑加權(quán)高差異最大,模型結(jié)果與坡度、郁閉度之間不存在趨勢性。
2.3 無人機(jī)激光雷達(dá)與攝影測量綜合分析森林結(jié)構(gòu)
無人機(jī)激光雷達(dá)對(duì)森林冠層具有穿透性,能夠準(zhǔn)確地獲取森林冠層內(nèi)部結(jié)構(gòu)及林下地形信息,但是獲取的光譜特征比較單一; 無人機(jī)攝影測量能夠獲取更為豐富的圖像光譜特征,通過圖像三維重建可以精細(xì)地觀測森林冠層上表面的空間分布特征,透過冠層之間的空隙獲取部分內(nèi)部結(jié)構(gòu)及林下地形信息,將無人機(jī)激光雷達(dá)和攝影測量相結(jié)合,可以彌補(bǔ)單一測量方式的不足 (Liseinetal., 2013; Wallaceetal., 2016)。
無人機(jī)激光雷達(dá)與攝影測量的結(jié)合方式一般是由激光雷達(dá)提取地形,將攝影測量點(diǎn)云減去激光雷達(dá)地形得到攝影測量歸一化點(diǎn)云,再提取森林結(jié)構(gòu)相關(guān)信息。Lisein等(2013) 使用固定翼無人機(jī)近紅外相機(jī)獲取橡樹(Quercuspalustris)林和云杉(Piceaasperata)林航空圖像,通過MICMAC提取點(diǎn)云數(shù)據(jù),由圖像重建DSM減去激光雷達(dá)DEM得到圖像重建CHM,與激光雷達(dá)CHM進(jìn)行對(duì)比分析,發(fā)現(xiàn)圖像重建CHM能夠很好地估測林分和單木變量; 許子乾等 (2015) 分析了固定翼無人機(jī)獲取的亞熱帶森林航空圖像,使用Pix4D進(jìn)行三維重建,并結(jié)合激光雷達(dá)DEM估測林分特征,發(fā)現(xiàn)胸徑加權(quán)高的敏感性最高,蓄積量次之,林分密度和胸高斷面積最低。另外,無人機(jī)激光雷達(dá)點(diǎn)云和攝影測量點(diǎn)云也可以合成在一起,再用激光雷達(dá)點(diǎn)云處理方法提取森林結(jié)構(gòu)相關(guān)信息。
無人機(jī)激光雷達(dá)和攝影測量容易受森林郁閉度的影響,Wallace等(2016) 使用多旋翼無人機(jī)激光雷達(dá)和相機(jī)集成系統(tǒng)同步獲取了桉樹林激光雷達(dá)點(diǎn)云和圖像數(shù)據(jù),采用Agisoft Photoscan從圖像重建三維點(diǎn)云,對(duì)比分析無人機(jī)激光雷達(dá)和攝影測量獲取的森林結(jié)構(gòu),結(jié)果發(fā)現(xiàn)對(duì)于低郁閉度區(qū)域,2種方法均能夠提供地形和冠層特征信息,隨著郁閉度增加,激光雷達(dá)能夠得到更高精度的垂直結(jié)構(gòu)信息。
森林變化主要指在一定時(shí)間間隔內(nèi),由于森林自身的生長演替、外部干擾等因素引起的森林結(jié)構(gòu)、組成成分等變化,多時(shí)相無人機(jī)激光雷達(dá)與攝影測量可以用于監(jiān)測多種形式的森林變化,包括森林物候變化、人工干擾引起的結(jié)構(gòu)變化等。由于無人機(jī)激光雷達(dá)獲取成本要高于攝影測量,一般采用無人機(jī)或有人機(jī)激光雷達(dá)獲取不易變化的林下地形信息,采用無人機(jī)攝影測量獲取多期森林結(jié)構(gòu)信息,2種技術(shù)相結(jié)合實(shí)現(xiàn)森林變化監(jiān)測。對(duì)于森林冠層的內(nèi)部結(jié)構(gòu)變化監(jiān)測,無人機(jī)攝影測量通常不能準(zhǔn)確地觀測這種變化,需要采用多時(shí)相無人機(jī)激光雷達(dá)進(jìn)行監(jiān)測。
3.1 森林物候變化
森林物候變化具有季節(jié)性,如冬季落葉期、早春/春季變綠期、夏季旺盛期、早秋衰變未落葉期和秋季衰落期等,對(duì)于落葉闊葉林或針葉林來說,不同季節(jié)枝條上葉片大小和葉量多少將直接影響激光雷達(dá)和攝影測量的點(diǎn)云密度,并影響森林結(jié)構(gòu)因子的估測精度。Dandois等(2013) 使用多旋翼無人機(jī)可見光相機(jī)獲取落葉林生長季和落葉季的航空圖像,飛行平臺(tái)距離樹冠約40 m,通過Agisoft Photoscan重建三維點(diǎn)云 (點(diǎn)云密度30~67點(diǎn)·m-2),分析不同季節(jié)圖像重建點(diǎn)云提取的DEM對(duì)森林結(jié)構(gòu)的影響,發(fā)現(xiàn)生長季圖像重建點(diǎn)云使用落葉季圖像重建DEM和激光雷達(dá)DEM得到樹高精度都很高,使用生長季圖像重建DEM的樹高精度要低一些; 另外,還選擇6種典型物候期進(jìn)行觀測,發(fā)現(xiàn)冠層相對(duì)綠度與MODIS NDVI時(shí)間序列高度相關(guān)。
3.2 森林結(jié)構(gòu)變化
多時(shí)相無人機(jī)激光雷達(dá)能夠準(zhǔn)確地測量人工干擾引起的森林結(jié)構(gòu)變化,如人工修枝、擇伐等森林經(jīng)營管理措施,進(jìn)而用于分析森林生物量、蓄積量等因子變化。Jaakkola等(2010) 選擇蘇格蘭松(Pinussylvestris)進(jìn)行多時(shí)相激光雷達(dá)試驗(yàn),分析枝葉的生物量變化,試驗(yàn)過程分為7個(gè)階段,每個(gè)階段手工去除部分枝葉,并對(duì)去除的枝葉稱重,使用無人機(jī)激光雷達(dá)觀測去除枝葉后的松樹,將松樹回波點(diǎn)數(shù)與地面回波點(diǎn)數(shù)的比值作為預(yù)測變量,與減少的枝葉生物量建立關(guān)系,發(fā)現(xiàn)二者之間的相關(guān)性很高 (R2=0.92)。Wallace等(2014c) 通過無人機(jī)激光雷達(dá)分析4年生桉樹人工林修枝效果,按照5組比率對(duì)樣地中隨機(jī)選取的樹木進(jìn)行修枝,提取每個(gè)修枝階段的枝下高,探測結(jié)果出現(xiàn)波動(dòng)性,探測修枝率與實(shí)測修枝率相比在96%~125%之間。
3.3 森林災(zāi)害監(jiān)測
森林災(zāi)害是林火、病蟲害等外部因素引起的森林減少現(xiàn)象,無人機(jī)攝影測量能夠用于森林災(zāi)害監(jiān)測和輔助決策,通過圖像識(shí)別方法提取林火點(diǎn)、煙霧或火場范圍、受害木等。無人機(jī)在林火發(fā)生前用于監(jiān)測植被和評(píng)估水分脅迫和火險(xiǎn)等級(jí),起火過程中用于林火探測、火情確認(rèn)、定位和監(jiān)測,過火后用于估測過火區(qū)并評(píng)價(jià)林火影響 (Olleroetal., 2006)。Ambrosia等(2003) 介紹了一種用于災(zāi)害監(jiān)測的無人機(jī)多光譜熱掃描成像儀,在林火漫延時(shí)間范圍內(nèi)能夠近實(shí)時(shí)傳輸幾何校正后的圖像給災(zāi)害管理人員; 楊斌等 (2009) 實(shí)現(xiàn)了一種無人機(jī)遙感圖像中煙的識(shí)別方法; 馬瑞升等 (2012) 設(shè)計(jì)了一種無人機(jī)林火監(jiān)測系統(tǒng),對(duì)火場影像進(jìn)行聚類分析,實(shí)時(shí)識(shí)別煙霧特征; 張?jiān)龅?(2015) 通過無人機(jī)圖像進(jìn)行森林火災(zāi)監(jiān)測,篩選具有林火的圖像,采用支持向量機(jī)進(jìn)行識(shí)別火災(zāi)區(qū)域。另外,何誠等 (2014) 通過無人機(jī)攝像系統(tǒng)對(duì)林火點(diǎn)進(jìn)行定位,根據(jù)無人機(jī)搭載的RTK GPS和微波測距儀計(jì)算林火點(diǎn)位置,定位誤差10 m以內(nèi)。
為了監(jiān)測大范圍的林火或從互補(bǔ)視角獲取林火信息,可以將多個(gè)無人機(jī)進(jìn)行編隊(duì),協(xié)同獲取林火信息。Martínez-de Dios等(2006) 介紹了一種用于林火監(jiān)測的無人機(jī)編隊(duì),由2個(gè)直升機(jī)和1個(gè)飛艇組成,搭載了1個(gè)可見光微型相機(jī)、1個(gè)近紅外微型相機(jī)和1個(gè)林火傳感器,對(duì)于可見光和近紅外圖像采用不同的林火識(shí)別算法; Merino等(2005) 提出了一種適合多無人機(jī)編隊(duì)的協(xié)作林火探測方法,采用計(jì)算機(jī)視覺技術(shù)在近紅外和可見光圖像和其他數(shù)據(jù)中識(shí)別和定位林火; Merino等(2006) 提出了一種適合多無人機(jī)編隊(duì)的協(xié)作感知系統(tǒng),考慮了近紅外、可見光和火探測器3種傳感器,包括圖像分割、序列圖像穩(wěn)定性保持、地理校正、協(xié)作感知融合等功能,用于林火探測和監(jiān)測; Merino(2010) 進(jìn)一步介紹了用于林火監(jiān)測的無人機(jī)編隊(duì)系統(tǒng),其中包括決策系統(tǒng)和感知系統(tǒng),決策系統(tǒng)用于任務(wù)分配、任務(wù)規(guī)劃和協(xié)調(diào),感知系統(tǒng)用于集成林火傳感器信息,無人機(jī)搭載了近紅外和可見光相機(jī),通過基于訓(xùn)練的閾值法在線提取林火信息; Karma等(2015) 使用無人機(jī)和地面交通工具模擬了林火搜救場景,3種無人機(jī) (固定翼、直升機(jī)和多旋翼) 用于巡邏、火漫延監(jiān)測、新火點(diǎn)制圖、預(yù)警和地面工具配合等,通過空氣質(zhì)量監(jiān)測和危險(xiǎn)區(qū)搜尋可以顯著提高人員安全性,存在的缺陷為交通工具的耐熱性、大風(fēng)和湍流環(huán)境中的飛行能力、復(fù)雜地形環(huán)境中的通信中斷等。
無人機(jī)在森林病蟲害應(yīng)用方面,通常在獲取的航空圖像上采用人工判斷或圖像識(shí)別等方式提取受害木信息。如胡根生等 (2013; 2014) 通過無人機(jī)平臺(tái)搭載的2個(gè)光譜相機(jī)獲取了針葉林可見光和近紅外遙感圖像,采用改進(jìn)的加權(quán)支持向量數(shù)據(jù)描述 (WSVDD) 和改進(jìn)的加權(quán)小波支持向量數(shù)據(jù)描述 (WWSVDD) 識(shí)別病害松樹,與K近鄰和支持向量數(shù)據(jù)描述分類方法相比,WWSVDD方法準(zhǔn)確性更高。
激光雷達(dá)能夠穿透森林冠層直接測量林下地形,對(duì)于高精度大區(qū)域地形制圖來說具有無可比擬的優(yōu)勢,歐美一些國家開始使用激光雷達(dá)生產(chǎn)區(qū)域或全國DEM產(chǎn)品 (Mengetal., 2010)。攝影測量能夠精確地獲取裸露地表的高程信息,對(duì)于植被覆蓋區(qū)來說,由于植被遮擋顯著降低了地形的測量精度 (Whiteetal., 2013)。攝影測量提取地形信息時(shí),一般根據(jù)圖像三維重建算法生成攝影測量點(diǎn)云,采用類似于激光雷達(dá)點(diǎn)云的方法提取地形信息 (Dandoisetal., 2010; 2013; Whiteetal., 2013; 2015)。
4.1 無人機(jī)激光雷達(dá)林下地形
激光雷達(dá)點(diǎn)云提取地形信息時(shí)通常根據(jù)點(diǎn)云之間的空間位置關(guān)系,采用不同的點(diǎn)云分類算法將回波點(diǎn)分為地面點(diǎn)和非地面點(diǎn),由離散地面點(diǎn)內(nèi)插生成DEM柵格。Meng等(2010) 將區(qū)分地面點(diǎn)和非地面點(diǎn)的地面濾波算法分為6類,即分割/聚類、形態(tài)學(xué)、方向掃描、等高線、TIN和插值,選擇地面濾波算法時(shí)應(yīng)考慮最優(yōu)回波 (首回波、中間回波、末回波) 和局部鄰域等因素。激光雷達(dá)點(diǎn)云提取地面點(diǎn)常用算法是漸近TIN密化算法 (Hyyppaetal., 2008; Wallaceetal., 2014a),基本思路是根據(jù)點(diǎn)云高程局部最小值生成初始地面點(diǎn)TIN表面模型,反復(fù)迭代搜索位于角度和距離閾值范圍內(nèi)的地面點(diǎn),更新地面點(diǎn)TIN表面模型 (Axelsson, 1999; 2000)。Wallace等(2014a; 2014b) 采用漸近TIN密化算法從無人機(jī)激光雷達(dá)點(diǎn)云中區(qū)分地面點(diǎn),地面濾波時(shí)僅使用了點(diǎn)云數(shù)據(jù)中的單次和末次回波,多次回波中的首次和中間回波被分為非地面點(diǎn),由自然鄰域內(nèi)插算法生成DEM柵格,通過重復(fù)性飛行試驗(yàn)對(duì)比分析,發(fā)現(xiàn)不同數(shù)據(jù)集DEM之間的差值不大(<0.39 m)。
4.2 無人機(jī)攝影測量林下地形
無人機(jī)攝影測量提取林下地形時(shí)需要根據(jù)圖像三維重建算法生成離散點(diǎn)云,由地面濾波算法識(shí)別地面點(diǎn)并內(nèi)插生成DEM柵格。Dandois等(2010) 使用異齡林和同齡林航空圖像進(jìn)行三維重建,通過圖像重建點(diǎn)云DEM與激光雷達(dá)DEM進(jìn)行對(duì)比,發(fā)現(xiàn)由于森林冠層遮擋影響,圖像重建DEM精度偏低; Dandois等(2013) 使用落葉林生長季和落葉季的航空圖像進(jìn)行三維重建,通過生長季和落葉季圖像重建DEM與激光雷達(dá)DEM進(jìn)行對(duì)比,發(fā)現(xiàn)落葉季圖像重建DEM精度 (RMSE為0.89~3.04 m) 比生長季圖像重建DEM精度 (RMSE為2.49~5.69 m) 高一些,對(duì)于圖像重建DEM與激光雷達(dá)DEM之間的差值來說,森林覆蓋區(qū)比非森林區(qū)域的DEM差值大; Wallace等(2016) 使用桉樹林航空圖像進(jìn)行三維重建,圖像重建DEM和激光雷達(dá)DEM總體上差異不大 (差值平均為0.09 m),對(duì)于高郁閉度區(qū)域,圖像重建DEM不如激光雷達(dá)DEM精確。
無人機(jī)激光雷達(dá)和攝影測量技術(shù)在林業(yè)中已經(jīng)開展了一些試驗(yàn)研究,用于測量單木/林分尺度的結(jié)構(gòu)特征,監(jiān)測森林恢復(fù)、森林經(jīng)營管理活動(dòng)等,但是不同的無人機(jī)平臺(tái)、傳感器數(shù)據(jù)獲取能力、數(shù)據(jù)采集模式、森林類型、森林物候特征、地形變化、光照條件等復(fù)雜因素均會(huì)影響森林信息提取精度,對(duì)于林業(yè)生產(chǎn)經(jīng)營管理業(yè)務(wù)來說,尚未形成適合推廣的成熟技術(shù)體系。
無人機(jī)激光雷達(dá)能夠直接測量森林結(jié)構(gòu)和林下地形,測量精度受激光采樣密度、幾何定位精度、航帶匹配精度、數(shù)據(jù)處理算法等因素影響; 多時(shí)相激光雷達(dá)測量森林結(jié)構(gòu)變化的誤差還受不同時(shí)相激光雷達(dá)數(shù)據(jù)配準(zhǔn)誤差的影響。激光雷達(dá)探測常綠森林時(shí),激光脈沖與森林冠層中的枝葉相互作用,葉子有效增加了后向散射截面,可以很精確地測量樹冠上層的結(jié)構(gòu)特征; 激光雷達(dá)探測落葉森林時(shí),由于樹冠內(nèi)枝干的有效后向散射面積多小于葉簇的相應(yīng)面積,導(dǎo)致枝干后向散射信號(hào)太弱,不能被激光雷達(dá)探測到,主要為來自地面的后向散射能量。
無人機(jī)攝影測量通過圖像三維重建點(diǎn)云來實(shí)現(xiàn)對(duì)森林空間結(jié)構(gòu)的測量,測量精度受圖像重疊率、圖像分辨率、匹配算法、點(diǎn)云密度、幾何定位精度、地形起伏等因素的影響 (St-Ongeetal., 2008; Hirshmuller, 2008; Whiteetal., 2013)。由于樹冠之間相互遮擋的影響,圖像三維重建僅能得到冠層表面的高度信息,很難得到林下地形信息,通常結(jié)合激光雷達(dá)獲取的林下地形得到攝影測量冠層高度。對(duì)于常綠森林來說,攝影測量通過圖像匹配得到冠層上表面枝葉的同名點(diǎn),冠層內(nèi)部和林下地表的同名點(diǎn)很少; 對(duì)于落葉森林來說,枝干對(duì)地面的遮擋很少,比較容易得到來自地面的同名點(diǎn),來自枝干同名點(diǎn)的數(shù)量依賴于圖像分辨率。
無人機(jī)激光雷達(dá)與攝影測量相結(jié)合可以彌補(bǔ)相互之間的不足,對(duì)于多時(shí)相森林動(dòng)態(tài)變化監(jiān)測、植被生長狀況監(jiān)測等長時(shí)間序列測量任務(wù)來說,考慮到林下地形很少變化,不同時(shí)間的森林結(jié)構(gòu)變化較大,且激光雷達(dá)的獲取成本要高于光學(xué)圖像,采用激光雷達(dá)獲取林下地形,攝影測量獲取時(shí)間序列森林表層結(jié)構(gòu)信息,保證激光雷達(dá)與攝影測量數(shù)據(jù)位置匹配精度的情況下,分析森林結(jié)構(gòu)變化規(guī)律。
隨著無人機(jī)系統(tǒng)負(fù)載、航時(shí)、穩(wěn)定性等性能進(jìn)一步提高,激光雷達(dá)、高精度GNSS & IMU、光學(xué)相機(jī)等遙感載荷的微型化、低成本化,以及相關(guān)數(shù)據(jù)處理軟件自動(dòng)化程度的提高,將使得無人機(jī)遙感系統(tǒng)在林業(yè)中的應(yīng)用范圍更為廣闊,以期服務(wù)于國家森林資源調(diào)查、重大工程監(jiān)測、生態(tài)環(huán)境監(jiān)測等業(yè)務(wù)需求。
崔紅霞, 林宗堅(jiān), 孫 杰. 2005. 無人機(jī)遙感監(jiān)測系統(tǒng)研究. 測繪通報(bào), (5): 11-14.
(Cui H X, Lin Z J, Sun J. 2005. Research on UAV remote sensing system. Bulletin of Surveying and Mapping, (5): 11-14.[in Chinese])
狄穎辰, 陳云坪, 陳瑩瑩, 等. 2011. 無人機(jī)圖像拼接算法綜述. 計(jì)算機(jī)應(yīng)用, 31(1): 170-174.
(Di Y C, Chen Y P, Chen Y Y,etal. 2011. Survey on image mosaic algorithm of unmanned aerial vehicle. Journal of Computer Applications, 31(1): 170-174. [in Chinese])
范承嘯, 韓 俊, 熊志軍, 等. 2009. 無人機(jī)遙感技術(shù)現(xiàn)狀與應(yīng)用. 測繪科學(xué), 34(5): 214-215.
(Fan C X, Han J, Xiong Z J,etal. 2009. Application and status of unmanned aerial vehicle remote sensing technology. Science of Surveying and Mapping, 34(5): 214-215. [in Chinese])
何 誠, 張思玉, 姚樹人. 2014. 旋翼無人機(jī)林火點(diǎn)定位技術(shù)研究. 測繪通報(bào), (12): 24-27.
(He C, Zhang S Y, Yao S R. 2014. Forest fires locating technology based on rotor UAV. Bulletin of Surveying and Mapping, (12): 24-27.[in Chinese])
洪 宇, 龔建華, 胡社榮, 等. 2008. 無人機(jī)遙感影像獲取及后續(xù)處理探討. 遙感技術(shù)與應(yīng)用, 23(4): 462-466.
(Hong Y, Gong J H, Hu S R,etal. 2008. Discussion on UAV remote sensing image capture and disposal. Remote Sensing Technology and Application, 23(4): 462-466. [in Chinese])
洪運(yùn)富, 楊海軍, 李 營, 等. 2015. 水源地污染源無人機(jī)遙感監(jiān)測. 中國環(huán)境監(jiān)測, 31(5): 163-166.
(Hong Y F, Yang H J, Li Y,etal. 2015. Monitoring of water source using unmanned aerial vehicle remote sensing technology. Environmental Monitoring in China, 31(5): 163-166. [in Chinese])
胡根生, 張學(xué)敏, 梁 棟, 等. 2013. 基于加權(quán)支持向量數(shù)據(jù)描述的遙感圖像病害松樹識(shí)別. 農(nóng)業(yè)機(jī)械學(xué)報(bào), 44(5): 258-263.
(Hu G S, Zhang X M, Liang D,etal. 2013. Infected pine recognition in remote sensing images based on weighted support vector data description. Transactions of the Chinese Society for Agricultural Machinery, 44(5): 258-263. [in Chinese])
胡根生, 張學(xué)敏, 梁 棟. 2014. 基于WWSVDD 多分類的遙感圖像病害松樹識(shí)別. 北京郵電大學(xué)學(xué)報(bào), 37(2): 23-27.
(Hu G S, Zhang X M, Liang D. 2014. Infected pine recognition in remote sensing images using WWSVDD multi-classification. Journal of Beijing University of Posts and Telecommunications, 37(2): 23-27. [in Chinese])
賈建軍, 舒 嶸, 王斌永. 2006. 無人機(jī)大面陣CCD相機(jī)遙感系統(tǒng). 光電工程, 33(8): 90-93.
(Jia J J, Shu R, Wang B Y. 2006. Large frame CCD camera system for unmanned aircraft remote sensing.Opto-Electronic Engineering, 33(8): 90-93. [in Chinese])
金 偉, 葛宏立, 杜華強(qiáng), 等. 2009. 無人機(jī)遙感發(fā)展與應(yīng)用概況. 遙感信息, (1): 88-92.
(Jin W, Ge H L, Du H Q,etal. 2009. A review on unmanned aerial vehicle remote sensing and its application. Remote Sensing Information, (1): 88-92. [in Chinese])
李德仁, 李 明. 2014. 無人機(jī)遙感系統(tǒng)的研究進(jìn)展與應(yīng)用前景. 武漢大學(xué)學(xué)報(bào):信息科學(xué)版, 39(5): 505-513, 540.
(Li D R, Li M. 2014. Research advance and application prospect of unmanned aerial vehicle remote sensing system. Geomatics and Information Science of Wuhan University, 39(5): 505-513, 540.[in Chinese])
李宇昊. 2007. 無人機(jī)在林業(yè)調(diào)查中的應(yīng)用實(shí)驗(yàn). 林業(yè)資源管理, (4): 69-73.
(Li Y H. 2007. Experiment on using unmanned aerial vehicle in forest investigation. Forest Resources Management, (4): 69-73.[in Chinese])
劉清旺, 李增元, 陳爾學(xué), 等. 2010. 機(jī)載LIDAR點(diǎn)云數(shù)據(jù)估測單株木生物量. 高技術(shù)通訊, 20(7): 765-770.
(Liu Q W, Li Z Y, Chen E X,etal. 2010. Estimating biomass of individual trees using point cloud data of airborne LIDAR. Chinese High Technology Letters, 20(7): 765-770. [in Chinese])
劉清旺, 李增元, 陳爾學(xué), 等. 2008. 利用機(jī)載激光雷達(dá)數(shù)據(jù)提取單株木樹高和樹冠. 北京林業(yè)大學(xué)學(xué)報(bào), 30(6): 83-89.
(Liu Q W, Li Z Y, Chen E X,etal. 2008. Extracting individual tree heights and crowns using airborne LIDAR data. Journal of Beijing Forestry University, 30(6): 83-89. [in Chinese])
呂書強(qiáng), 晏 磊, 張 兵, 等. 2007. 無人機(jī)遙感系統(tǒng)的集成與飛行試驗(yàn)研究. 測繪科學(xué), 32(1): 84-86.
(Lü S Q, Yan L, Zhang B,etal. 2007. The integration and flight experiment of UAV remote sensing system. Science of Surveying and Mapping, 32(1): 84-86 [in Chinese])
馬瑞升, 楊 斌, 張利輝, 等. 2012. 微型無人機(jī)林火監(jiān)測系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn). 浙江農(nóng)林大學(xué)學(xué)報(bào), 29(5): 783-789.
(Ma R S, Yang B, Zhang L H,etal. 2012. Design and implementation of a forest fire monitoring system using a miniature unmanned aerial vehicle. Journal of Zhejiang A&F University, 29(5): 783-789. [in Chinese])
秦 博, 王 蕾. 2002. 無人機(jī)發(fā)展綜述. 飛航導(dǎo)彈, (8): 4-10.
(Qin B, Wang L. 2002. Review of the development of UAV. Aerodynamic Missile Journal, (8): 4-10.[in Chinese])
孫 杰, 林宗堅(jiān), 崔紅霞. 2003. 無人機(jī)低空遙感監(jiān)測系統(tǒng). 遙感信息, (1): 49-50, 27.
(Sun J, Lin Z J, Cui H X. 2003. Alow-altitude UAV remote sensing monitoring system. Remote Sensing Information, (1): 49-50, 27. [in Chinese])
汪 沛, 羅錫文, 周志艷, 等. 2014. 基于微小型無人機(jī)的遙感信息獲取關(guān)鍵技術(shù)綜述. 農(nóng)業(yè)工程學(xué)報(bào), 30(18): 1-12.
(Wang P, Luo X W, Zhou Z Y,etal. 2014. Key technology for remote sensing information acquisition based on micro UAV. Transactions of the Chinese Society of Agricultural Engineering, 30(18): 1-12. [in Chinese])
王 峰, 吳云東. 2010. 無人機(jī)遙感平臺(tái)技術(shù)研究與應(yīng)用. 遙感信息, (2): 114-118.
(Wang F, Wu Y D. 2010. Research and application of UAS borne remote sensing. Remote Sensing Information, (2): 114-118.[in Chinese])
王利民, 劉 佳, 楊玲波, 等. 2013. 基于無人機(jī)影像的農(nóng)情遙感監(jiān)測應(yīng)用. 農(nóng)業(yè)工程學(xué)報(bào), 29(18): 136-145.
(Wang L M, Liu J, Yang L B,etal. 2013. Applications of unmanned aerial vehicle images on agricultural remote sensing monitoring. Transactions of the Chinese Society of Agricultural Engineering, 29(18): 136-145. [in Chinese])
許子乾, 曹 林, 阮宏華, 等. 2015. 集成高分辨率UAV影像與激光雷達(dá)點(diǎn)云的亞熱帶森林林分特征反演. 植物生態(tài)學(xué)報(bào), 39(7): 694-703.
(Xu Z Q, Cao L, Ruan H H,etal. 2015. Inversion of subtropical forest stand characteristics by integrating very high resolution imagery acquired from UAV and LiDAR point-cloud. Chinese Journal of Plant Ecology, 39(7): 694-703. [in Chinese])
楊 斌, 馬瑞升, 何 立, 等. 2009. 基于顏色特征的遙感圖像中煙的識(shí)別方法. 計(jì)算機(jī)工程, 35(7): 168-169.
(Yang B, Ma R S, He L,etal. 2009. Method of identifying fume in remote sensing images based on color features. Computer Engineering, 35(7): 168-169. [in Chinese])
楊海軍, 李 營, 朱海濤, 等. 2015. 無人機(jī)遙感技術(shù)在環(huán)境保護(hù)領(lǐng)域的應(yīng)用. 高技術(shù)通訊, 25(6): 607-613.
(Yang H J, Li Y, Zhu H T,etal. 2015. UAV remote sensing’s applications in the environmental protection field. High Technology Letters, 25(6): 607-613. [in Chinese])
張 波, 羅錫文, 蘭玉彬, 等. 2015. 基于無線傳感器網(wǎng)絡(luò)的無人機(jī)農(nóng)田信息監(jiān)測系統(tǒng). 農(nóng)業(yè)工程學(xué)報(bào), 31(17): 176-182.
(Zhang B, Luo X W, Lan Y B,etal. 2015. Agricultural environment monitor system based on UAV and wireless sensor networks. Transactions of the Chinese Society of Agricultural Engineering, 31(17): 176-182. [in Chinese])
張 增, 王 兵, 伍小潔, 等. 2015. 無人機(jī)森火災(zāi)監(jiān)測中火情檢測方法研究. 遙感信息, 30(1): 107-110, 124.
(Zhang Z, Wang B, Wu X J,etal. 2015. An algorithm of forest fire detection based on UAV remote sensing. Remote Sensing Information, 30(1): 107-110, 124. [in Chinese])
周潔萍, 龔建華, 王 濤, 等. 2008. 汶川地震災(zāi)區(qū)無人機(jī)遙感影像獲取與可視化管理系統(tǒng)研究. 遙感學(xué)報(bào), 12(6): 877-884.
(Zhou J P, Gong J H, Wang T,etal. 2008. Study on UAV remote sensing image acquiring and visualization management system for the area affected by 5·12 Wenchuan Earthquake. Journal of Remote Sensing, 12(6): 877-884. [in Chinese])
朱 鋒, 肖 暉, 魏亞男. 2014. 無人機(jī)遙感影像鑲嵌技術(shù)綜述. 計(jì)算機(jī)工程與應(yīng)用, 50(15): 38-41.
(Zhu F, Xiao H, Wei Y N. 2014. Summary of UAV remote sensing image mosaicking technology. Computer Engineering and Applications, 50(15): 38-41.[in Chinese])
Alexander C. 2009. Delineating tree crowns from airborne laser scanning point cloud data using Delaunay triangulation. International Journal of Remote Sensing, 30(14): 3843-3848.
Ambrosia V, Wegener S, Sullivan D,etal. 2003. Demonstrating UAV-aquired real-time thermal data over fires. Photogrametric Engineering and Remote Sensing, 69(4): 391-402.
Amon P, Riegl U, Rieger P,etal. 2015. UAV-based laser scanning to meet special challenges in lidar surveying. Geomatics Indaba Proceedings 2015, 138-147.
Axelsson P. 1999. Processing of laser scannerdata-algorithms and applications. ISPRS Journal of Photogrammetry and Remote Sensing, 54(2/3): 138-147.
Axelsson P. 2000. DEM generation from laser scanner data using adaptive TIN models. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XIX ISPRS Congress,XXXIII-B4, 110-117.
Bendig J, Bolten A, Bareth G. 2012. Introducing alow-cost mini-UAV for thermal and multispectral-imaging. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXII ISPRS Congress, XXXIX-B1, 345-349.
Berman E S F, Fladeland M, Liem J,etal. 2012. Greenhouse gas analyzer for measurements of carbon dioxide, methane, and water vapor aboard an unmanned aerial vehicle. Sensors and Actuators B: Chemical, 169: 128-135.
Brandtberg T, Warner T A, Landenberger R E,etal. 2003. 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. Remote Sensing of Environment, 85(3): 290-303.
Brandtberg T. 2007. Classifying individual tree species underleaf-off and leaf-on conditions using airborne lidar. ISPRS Journal of Photogrammetry and Remote Sensing, 61(5): 325-340.
Breckenridge R P, Dakins M, Bunting S,etal. 2011. Comparison of unmanned aerial vehicle platforms for assessing vegetation cover in sagebrush steppe ecosystems. Rangeland Ecology & Management, 64(5): 521-532.
Breckenridge R P, Dakins M, Bunting S,etal. 2012. Using unmanned helicopters to assess vegetation cover in sagebrush steppe ecosystems. Rangeland Ecology & Management, 65(4): 362-370.
Choi K, Lee I, Hong J,etal. 2009. Developing aUAV-based rapid mapping system for emergency response. Proceedings of SPIE 7332, Unmanned Systems Technology XI, 733209.
Colomina I, Molina P. 2014. Unmanned aerial systems for photogrammetry and remote sensing: a review. ISPRS Journal of Photogrammetry and Remote Sensing, 92: 79-97.
Córcoles J I, Ortega J F, Hernández D,etal. 2013. Estimation of leaf area index in onion (AlliumcepaL.) using an unmanned aerial vehicle. Biosystems Engineering, 115(1): 31-42.
Dandois J P, Ellis E C. 2013. High spatial resolutionthree-dimensional mapping of vegetation spectral dynamics using computer vision. Remote Sensing of Environment, 136: 259-276.
Dandois J P, Ellis E C. 2010. Remote sensing of vegetation structure using computer vision. Remote Sensing, 2(4): 1157-1176.
Dandois J P, Olano M, Ellis E C. 2015. Optimal altitude, overlap, and weather conditions for computer vision UAV estimates of forest structure. Remote Sensing, 7(10): 13895-13920.
Díaz-Varela R A, Rosa R, León L,etal. 2015. High-resolution airborne UAV imagery to assess olive tree crown parameters using 3D photo reconstruction: application in breeding trials. Remote Sensing, 7(4): 4213-4232.
Ditmer M A, Vincent J B, Werden L K,etal. 2015. Bears show a physiological but limited behavioral response to unmanned aerial vehicles. Current Biology, 25 (17): 2278-2283.
Dunford R, Michel K, Gagnage M,etal. 2009. Potential and constraints of unmanned aerial vehicle technology for the characterization of mediterranean riparian forest. International Journal of Remote Sensing, 30(19): 4915-4935.
Essen H, Johannes W, Stanko S,etal. 2012. High resolution W-band UAV SAR. IEEE International Geoscience and Remote Sensing Symposium (IGARSS) 2012, 5033-5036.
Floreano D, Wood R J. 2015. Science, technology and the future of small autonomous drones. Nature, 521(7553): 460-466.
García-Gutiérrez J, Martínez-lvarez F, Troncoso A,etal. 2015. A comparison of machine learning regression techniques for LiDAR-derived estimation of forest variables. Neurocomputing, 167: 24-31.
Gehrke S, Downey M, Uebbing R,etal. 2012. A multi-sensor approach to semi-global matching. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXII ISPRS Congress, XXXIX-B3, 17-22.
Gehrke S, Uebbing R, Downey M,etal. 2011. Creating and using very high density point clouds derived from ADS imagery. Proceedings of the American Society of Photogrammetry and Remote Sensing 2011 Annual Conference.
Getzin S, Nuske R S, Wiegand K. 2014. Using unmanned aerial vehicles (UAV) to quantify spatial gap patterns in forests. Remote Sensing, 6(8): 6988-7004.
Ginzler C, Hobi M. 2015. Countrywidestereo-image matching for updating digital surface models in the framework of the Swiss national forest inventory. Remote Sensing, 7(4): 4343-4370.
Gleasona C J, Im J. 2012. Forest biomass estimation from airborne LiDAR data using machine learning approaches. Remote Sensing of Environment, 125: 80-91.
Glennie C, Brooks B, Ericksen T,etal. 2013. Compact multipurpose mobile laser scanning system — initial tests and results. Remote Sensing, 5(2): 521-538.
Goodwin N, Coops N, Culvenor D. 2006. Assessment of forest structure with airborne LiDAR and the effects of platform altitude. Remote Sensing Environment, 103 (2): 140-152.
Gottfried M, Hollaus M, Glira P,etal. 2015. First examples from the RIEGL VUX-SYS for forestry applications. In Proceedings of SilviLaser 2015, 105-107.
Grenzd?rffer G, Niemeyer F, Schmidt F. 2012. Development of four vision camera system for amicro-UAV. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXII ISPRS Congress, XXXIX-B1, 369-374.
Gupta S, Weinacker H, Koch B. 2010. Comparative analysis ofclustering-based approaches for 3-D single tree detection using airborne fullwave LiDAR data. Remote Sensing, 2 (4): 968-989.
Hernández-Clemente R, Navarro-Cerrillo R M, Zarco-Tejad P J. 2012. Carotenoid content estimation in a heterogeneous conifer forest using narrow-band indices and PROSPECT + DART simulations. Remote Sensing of Environment, 127: 298-315.
Hirshmuller H. 2008. Stereo processing bysemi-global matching and mutual information. IEEE Transactions on Pattern Analysis and Machine Intelligence, 30(2): 328-341.
Holmgren J. 2004. Prediction of tree height, basal area and stem volume in forest stands using airborne laser scanning. Scandinavian Journal of Forest Research, 19(6): 543-553.
Hudak A T, Crookston N L, Evans J S,etal. 2008. Nearest neighbor imputation of species-level, plot-scale forest structure attributes from LiDAR data. Remote Sensing of Environment, 112(5): 2232-2245.
Hyyppa J, Hyyppa H, Leckie D,etal. 2008. Review of methods of small-footprint airborne laser scanning for extracting forest inventory data in boreal forests. International Journal of Remote Sensing, 29(5): 1339-1366.
Immerzeel W W, Kraaijenbrink P D A, Shea J M,etal. 2014. High-resolution monitoring of Himalayan glacier dynamics using unmanned aerial vehicles. Remote Sensing of Environment, 150: 93-103.
Jaakkola A, Hyypp? J, Kukko A,etal. 2010. A low-cost multi-sensoral mobile mapping system and its feasibility for tree measurements. ISPRS Journal of Photogrammetry and Remote Sensing, 65(6): 514-522.
Kaartinen H, Hyypp? J, Yu X,etal. 2012. An international comparison of individual tree detection and extraction using airborne laser scanning. Remote Sensing, 4 (4): 950-974.
Karma S, Zorba E, Pallis G C,etal. 2015. Use of unmanned vehicles in search and rescue operations in forest fires: advantages and limitations observed in a field trial. International Journal of Disaster Risk Reduction, 13: 307-312.
Kelcey J., Lucieer A. 2012. Sensor correction and radiometric calibration of a 6-band multispectral imaging sensor for UAV remote sensing. In International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXII ISPRS Congress,XXXIX-B1, 393-398.
Koch B, Heyder U, Weinacker H.2006. Detection of individual tree crowns in airborne Lidar data. Photogrammetric Engineering & Remote Sensing, 72(4): 357-363.
L?hivaara T, Sepp?nen A, Kaipio J P,etal. 2012. Bayesian approach to tree detection with airborne laser scanning. IEEE International Geoscience and Remote Sensing Symposium (IGRSS) 2012, 1641-1644.
Lehmann J R K, Nieberding F, Prinz T,etal. 2015. Analysis of unmanned aerial system-based CIR images in forestry—a new perspective to monitor pest infestation levels. Forests, 6(3): 594-612.
Li W, Guo Q. 2012. A new method for segmenting individual trees from the LiDAR point cloud. Photogrammetric Engineering & Remote Sensing, 78(1): 75-84.
Lin Y, Hyypp? J, Jaakkola A. 2011.Mini-UAV-borne LIDAR for fine-scale mapping. IEEE Geoscience and Remote Sensing Letters, 8(3): 426-430.
Lisein J,Pierrot-Deseilligny M, Bonnet S,etal. 2013. A photogrammetric workflow for the creation of a forest canopy height model from small unmanned aerial system imagery. Forests, 4(4): 922-944.
Liu C C, Chen Y H, Wen H L. 2015. Supporting the annual international black-faced spoonbill census with a low-cost unmanned aerial vehicle. Ecological Informatics, 30: 170-178.
Liu Q, Li Z, Chen E,etal. 2011. Feature analysis of LIDAR waveforms from forest canopies. Science China Earth Sciences, 54(8): 1206-1214.
Maltamo M, Eerikainen K, Pitkanen J,etal. 2004. Estimation of timber volume and stem density based on scanning laser altimetry and expected tree size distribution functions. Remote Sensing of Environment, 90(3): 319-330.
Martínez-de Dios J R, Merino L, Caballero F,etal. 2006. Experimental results of automatic fire detection and monitoring with UAVs. Proceedings of V International Conference on Forest Fire Research.
Masek J G, Hayes D J, Hughes M J,etal. 2015. The role of remote sensing in process-scaling studies of managed forest ecosystems. Forest Ecology and Management, 355: 109-123.
McElhinny C, Gibbons P, Brack C,etal. 2005. Forest and woodland stand structural complexity: its definition and measurement. Forest Ecology and Management, 218(1/3): 1-24.
Meng X, Currit N, Zhao K. 2010. Ground filtering algorithms for airborne LiDAR data: a review of critical issues. Remote Sensing, 2(3): 833-860.
Merino L, Caballero F, Martínez de Dios J R,etal. 2006. A cooperative perception system for multiple UAVs: application to automatic detection of forest fires. Journal of Field Robotics, 23 (3/4): 165-184.
Merino L, Caballero F, Martínez de Dios J R,etal. 2005. Cooperative fire detection using unmanned aerial vehicles. In Proceedings of the IEEE International Conference on Robotics and Automation, 1896-1901.
Merino L. 2010. Automatic forest fire monitoring and measurement using unmanned aerial vehicles. VI International Conference on Forest Fire Research.
Mesas-Carrascosa F J, Notario-García M D, de Larriva J E M,etal. 2014. Validation of measurements of land plot area using UAV imagery. International Journal of Applied Earth Observation and Geoinformation, 33: 270-279.
Morsdorf F, Meier E, Allg?wer B,etal. 2003. Clustering in airborne laser scanning raw data for segmentation of single trees. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 3-D reconstruction from airborne laserscanner and InSAR data, XXXIV-3/W13, 27-33.
Nagai M, Shibasaki R, Manandhar D,etal. 2004. Development of digital surface and feature extraction by integrating laser scanner and CCD sensor with IMU. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XX ISPRS Congress, XXXV-B5, 655-659.
Nagai M, Tianen C, Shibasaki R,etal. 2009. UAV-borne 3-D mapping system by multisensor integration. IEEE Transactions on Geoscience and Remote Sensing, 47 (3): 701-708.
Ni W, Liu J, Zhang Z,etal. 2015. Evaluation of UAV-based forest inventory system compared with LiDAR data. IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 3874-3877.
Ollero A, Martínez-de-Dios J R, Merino L. 2006. Unmanned aerial vehicles as tools for forest-fire fighting. Proceedings of V International Conference on Forest Fire Research.
Ota T, Ogawa M, Shimizu K,etal. 2015. Aboveground biomass estimation using structure from motion approach with aerial photographs in a seasonal tropical forest. Forests, 6(11): 3882-3898.
Packalén P, Maltamo M. 2007. Thek-MSN method for the prediction of species-specific stand attributes using airborne laser scanning and aerial photographs. Remote Sensing of Environment, 109(3): 328-341.
Pajares G. 2015. Overview and current status of remote sensing applications based on unmanned aerial vehicles (UAVs). Photogrammetric Engineering & Remote Sensing, 81 (4): 281-329.
Penner M, Woods M, Pitt D A. 2015. Comparison of airborne laser scanning and image point cloud derived tree size class distribution models in Boreal Ontario. Forests, 6(11): 4034-4054.
Popescu S C. 2007. Estimating biomass of individual pine trees using airborne lidar. Biomass and Bioenergy, 31(9): 646-655.
Rehak M, Mabillard R, Skaloud J. 2013. Amicro-UAV with the capability of direct georeferencing. In International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, UAV-g 2013, XL-1/W2, 317-323.
Reitberger J, Schn?rr C, Krzystek P,etal. 2009. 3D segmentation of single trees exploiting full waveform LIDAR data. ISPRS Journal of Photogrammetry and Remote Sensing, 64(6): 561-574.
Rejas J I, Sanchez A, Glez-de-Riveraa G,etal. 2015. Environment mapping using a 3D laser scanner for unmanned ground vehicles. Microprocessors and Microsystems, 39 (8): 939-949.
Remy M A, de Macedo K A C, Moreira J R. 2012. The FirstUAV-based P- and X-band Interferometric SAR System. IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 5041-5044.
Rufino G, Moccia A. 2005. IntegratedVIS-NIR Hyperspectral/thermal-IR Electrooptical Payload System for a Mini-UAV. American Institute of Aeronautics and Astronautics, 647-664.
Scherer S, Singh S, Chamberlain L,etal. 2008. Flying fast and low among obstacles: methodology and experiments. International Journal of Robotics Research, 27(5): 549-574.
Scholtz A, Kaschwich C, Kruger A,etal. 2011. Development of a new multi-purpose UAS for scientific application. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, International Conference on Unmanned Aerial Vehicle in Geomatics (UAV-g), XXXVIII-1/C22, 149-154.
Shahbazi M, Sohn G, Théau J,etal. 2015. Development and evaluation of a UAV-photogrammetry system for precise 3D environmental modeling. Sensors, 15(11): 27493-27524.
Snavely N, Seitz S M, Szeliski R. 2008. Modeling the world from Internet photo collections. International Journal of Computer Vision, 80(2): 189-210.
St-Onge B, Audet F-A, Bégin J. 2015. Characterizing the height structure and composition of a boreal forest using an individual tree crown approach applied to photogrammetric point clouds. Forests, 6(11): 3899-3922.
St-Onge B, Vega C, Fournier R A,etal. 2008. Mapping canopy height using a combination of digital stereo-photogrammetry and lidar. International Journal of Remote Sensing, 29(11): 3343-3364.
Tulldahl H M, Bissmarck F, Larsson H,etal. 2015. Accuracy evaluation of 3D lidar data from small UAV. Proceedings of SPIE 9649, Electro-Optical Remote Sensing, Photonic Technologies, and Applications IX, 964903.
Tulldahl H. M, Larsson H. 2014. Lidar on small UAV for 3D mapping. Proceedings of SPIE 9250,Electro-Optical Remote Sensing, Photonic Technologies, and Applications VIII, and Military Applications in Hyperspectral Imaging and High Spatial Resolution Sensing II, 925009.
Turner D, Lucieer A, Watson C. 2012. An automated technique for generating georectified mosaics fromultra-high resolution unmanned aerial vehicle (UAV) imagery, based on structure from motion (SfM) point clouds. Remote Sensing, 4(5): 1392-1410.
Turner D, Lucieer A, Wallace L. 2014. Direct georeferencing of ultrahigh-resolution UAV imagery. IEEE Transactions on Geoscience and Remote Sensing, 52(5): 2738-2745.
Vaughn N R, Moskal L M, Turnblom E C. 2012. Tree species detection accuracies using discrete point LiDAR and airborne waveform LiDAR. Remote Sensing,4(2): 377-403.
Vauhkonen J, Ene L, Gupta S,etal. 2011. Comparative testing of single-tree detection algorithms under different types of forest. Forestry, 85(1): 27-40.
Wallace L, Lucieer A, Malenovsk Z,etal. 2016. Assessment of forest structure using two UAV techniques: a comparison of airborne laser scanning and structure from motion (SfM) point clouds. Forests, 7(3): 62.
Wallace L, Lucieer A, Watson C,etal. 2012. Development of a UAV-LiDAR system with application to forest inventory. Remote Sensing, 4(6): 1519-1543.
Wallace L, Lucieer A, Watson C S. 2014a. Evaluating tree detection and segmentation routines on very high resolution UAV LiDAR data. IEEE Transactions on Geoscience and Remote Sensing, 52(12): 7619-7628.
Wallace L, Musk R, Lucieer A. 2014b. An assessment of the repeatability of automatic forest inventory metrics derived fromUAV-borne laser scanning data. IEEE Transactions on Geoscience and Remote Sensing, 52(11): 7160-7169.
Wallace L, Watson C, Lucieer A. 2014c. Detecting pruning of individual stems using airborne laser scanning data captured from an Unmanned Aerial Vehicle. International Journal of Applied Earth Observation and Geoinformation, 30: 76-85.
Wang Y, Weinacker H, Koch B,etal. 2008. Lidar point cloud based fully automatic 3D single tree modelling in forest and evaluations of the procedure. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXI ISPRS Congress, XXXVII, 45-51.
Waser L T, Fischer C, Wang Z,etal. 2015. Wall-to-wall forest mapping based on digital surface models from image-based point clouds and a NFI forest definition. Forests, 6(12): 4510-4528.
Watts A C, Ambrosia V G, Hinkley E A. 2012. Unmanned aircraft systems in remote sensing and scientific research: classification and considerations of use. Remote Sensing, 4(6): 1671-1692.
White J C, Wulder M A, Vastaranta M,etal. 2013. The utility of image-based point clouds for forest inventory: a comparison with airborne laser scanning. Forests, 4(3): 518-536.
White J, Stepper C, Tompalski P,etal. 2015. Comparing ALS and image-based point cloud metrics and modelled forest inventory attributes in a complex coastal forest environment. Forests, 6(10): 3704-3732.
Xiang H., Tian L. 2011. Development of alow-cost agricultural remote sensing system based on an autonomous unmanned aerial vehicle (UAV). Biosystems Engineering, 108(2): 174-190.
Yang B, Chen C. 2015. Automatic registration ofUAV-borne sequent images and LiDAR data. ISPRS Journal of Photogrammetry and Remote Sensing, 101: 262-274.
Yu X, Hyypp? J, Vastaranta M,etal. 2011. Predicting individual tree attributes from airborne laser point clouds based on the random forests technique. ISPRS Journal of Photogrammetry and Remote Sensing, 66(1): 28-37.
Zahawi R A, Dandois J P, Holl K D,etal. 2015. Using lightweight unmanned aerial vehicles to monitor tropical forest recovery. Biological Conservation, 186: 287-295.
Zarco-Tejada P J, Guillén-Climent M L, Hernández-Clemente R,etal. 2013. Estimating leaf carotenoid content in vineyards using high resolution hyperspectral imagery acquired from an unmanned aerial vehicle (UAV). Agricultural and Forest Meteorology, 171- 172: 281- 294.
Zarco-Tejada P J, Diaz-Varela R, Angileri V,etal. 2014. Tree height quantification using very high resolution imagery acquired from an unmanned aerial vehicle (UAV) and automatic 3D photo-reconstruction methods. European Journal of Agronomy, 55: 89-99.
Zhang C, Kovacs J M. 2012. The application of small unmanned aerial systems for precision agriculture: A review. Precision Agriculture, 13: 693-712.
Zhang J, Hu J, Lian J,etal. 2016. Seeing the forest from drones: Testing the potential of lightweight drones as a tool for long-term forest monitoring. Biological Conservation, 198: 60-69.
Zhao K, Popescu S, Meng X,etal. 2011. Characterizing forest canopy structure with lidar composite metrics and machine learning. Remote Sensing of Environment, 115(8): 1978-1996.
Zhou G, Yang J, Li X,etal. 2012. Advances of flash LiDAR development onboard UAV. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXII ISPRS Congress, XXXIX-B3, 193-198.
(責(zé)任編輯 石紅青)
Review on the Applications of UAV-Based LiDAR and Photogrammetry in Forestry
Liu Qingwang1Li Shiming1Li Zengyuan1Fu Liyong1Hu Kailong1, 2
(1.ResearchInstituteofForestResourceInformationTechniques,CAFBeijing100091; 2.CollegeofGeo-ScienceandSurveyingEngineering,ChinaUniversityofMining&TechnologyBeijing100083)
Forest spatial structure and dynamics pattern are crucial to forest management and ecological modelling. Unmanned aerial vehicle (UAV) based light detecting and ranging (LiDAR) and photogrammetry could provide comprehensive spatial structure and species of forest, and have unrivalled advantages in the long-time monitoring of forest environment at individual tree or stand scale. UAV-based LiDAR system usually carries multiple echoes / full wave laser scanner, and assembles high precision global navigation satellite system (GNSS) & inertial measurement unit (IMU) which is used to ensure the position accuracy of backscatter signals of transmitted laser pulses. UAV-based photogrammetry system mainly carries visual (RGB) / multiband camera, and assembles low precision GNSS & IMU. Automated 3D reconstruction algorithms can estimate the locations and orientations of cameras and camera internal parameters using highly overlapping aerial photographs, and generate initial rectified images and point cloud with relative coordinates, which can be georeferenced by ground control points (GCPs), reference images, etc. The accuracy of image matching can be improved using high precision GNSS, stabilized platform, etc. Individual tree segmentation algorithms were generally used to extract structure information of individual trees, such as tree tops, crown edges, locations of trees, etc., from point cloud of LiDAR or photogrammetry reconstruction. The structure features of individual trees can also be recognized from projected voxel space or canopy height model (CHM) generated from point cloud. Forest stand structure information were usually estimated by height profile algorithms from point cloud or synthetic waveform. The point cloud can be directly used to calculate features, such as height percentile, echo index, etc., or generate synthetic waveforms based frequency or intensity of echoes at specified bin of height. The waveform features, such as percentile, leading edge, trailing edge, etc., can be extracted from synthetic waveforms. The estimation values of forest structure parameters were obtained based on the relationship between field measurements and the features of point cloud or waveforms. The terrain under forest canopy can be detected from point cloud of LiDAR or photogrammetry reconstruction. The accuracy of terrain from photogrammetry reconstruction was similar to that from LiDAR in low canopy closure area, but lower than that from LiDAR in high canopy closure area. Multitemporal measurements of UAV-based LiDAR and photogrammetry can be used to monitor forest structure change caused by manual pruning, selective cutting, forest fire, disease and pest damage, etc., and phenological change, such as brunches and leaves growing, leaves falling, etc. The estimation accuracy of forest structure parameters extracted using UAV-based LiDAR and photogrammetry were affected by acquisition patterns, data processing algorithms, forest growing season, terrain, etc. The art of state repertoire hasn’t been suitable to wide utilization in forestry. The UAV flying should follow the constrains of national / local laws and regulations, which has been managed according to some conditions, such as empty weight, max take-off weight, etc., in China. In the future, UAV data acquisition and processing system will be more intelligent, miniaturized, low-cost, and better serve the needs of forestry applications.
unmanned aerial vehicle(UAV); light detection and ranging(LiDAR); photogrammetry; point cloud; forest
10.11707/j.1001-7488.20170714
2016-04-13;
2016-07-06。
中國林業(yè)科學(xué)研究院中央級(jí)公益性科研院所基本科研業(yè)務(wù)費(fèi)專項(xiàng)資金項(xiàng)目(CAFYBB2016SZ003); 國家重點(diǎn)基礎(chǔ)研究發(fā)展計(jì)劃(973計(jì)劃)課題(2013CB733405, 2013CB733404)。
S757
A
1001-7488(2017)07-0134-15
* 李世明為通訊作者。