1、第32卷第1期2007年1月測(cè)繪科學(xué)Science of Surveying and M app ingVol 132No 11Jan 1 作者簡(jiǎn)介:李勤超(1982 ,男,湖北天門(mén)人,解放軍信息工程大學(xué)測(cè)繪學(xué)院六系三隊(duì)450052,碩士生,主要從事地理信息系統(tǒng)研究。E mail:poll_lenyahoo 1com 1cn收稿日期:20060316基于D E M 提取水域特征的一種算法實(shí)現(xiàn)李勤超,李宏偉,孟嬋媛(解放軍信息工程大學(xué)測(cè)繪學(xué)院,鄭州450052;海軍海洋測(cè)繪研究所,天津300061【摘要】本文提出并實(shí)現(xiàn)了一種基于規(guī)則格網(wǎng)DE M 的地形特征線提取算法。該算法基于水文模擬方法,提出

2、了平地、洼地的結(jié)構(gòu)模型,研究了它們形成復(fù)雜洼地的拓?fù)潢P(guān)系,在此基礎(chǔ)上設(shè)計(jì)了洼地填平等算法,簡(jiǎn)便快捷,在邙山小流域水土保持項(xiàng)目中得到應(yīng)用,效果較好。【關(guān)鍵字】格網(wǎng)DE M;水文模擬;地形特征;水域特征【中圖分類(lèi)號(hào)】P282【文獻(xiàn)標(biāo)識(shí)碼】A 【文章編號(hào)】10092307(2007010103021引言河流水系的提取技術(shù)被廣泛應(yīng)用于地圖制圖、水文模擬和地質(zhì)災(zāi)害預(yù)測(cè)等眾多研究領(lǐng)域。傳統(tǒng)的人工方法不僅費(fèi)時(shí)費(fèi)力,而且由于水系級(jí)別的人為確定和低級(jí)別水流線的刪除等問(wèn)題,造成了許多誤差。近年來(lái),隨著地理信息系統(tǒng)(GI S 的日益普及,特別是能夠表達(dá)豐富地貌信息的數(shù)字高程模型(DE M 技術(shù)的出現(xiàn),使從DE M

3、中提取地貌特征信息成為一項(xiàng)重要的研究工作。中外學(xué)者提出了許多自動(dòng)提取的方法,并開(kāi)發(fā)了相關(guān)的軟件。根據(jù)算法設(shè)計(jì)的原理,從DE M 中自動(dòng)提取水系網(wǎng)絡(luò)的方法主要分為三類(lèi)1:基于地形表面幾何形態(tài)的方法,也稱(chēng)之為局部方法,如Johnst on 和Rosenfeld 等的地形方向斷面極值法,以及Tribe 的河谷網(wǎng)絡(luò)算法,該方法提取的河流網(wǎng)絡(luò),存在河流結(jié)點(diǎn)的不連續(xù)性以及河流源頭定位的隨意性等問(wèn)題;基于地表水流模擬的方法,也稱(chēng)之為全局方法,它首先給每個(gè)DE M 單元賦流向,然后根據(jù)所有單元的流向產(chǎn)生一個(gè)全連接的、單個(gè)單元寬度的流向網(wǎng)絡(luò),最后根據(jù)集水區(qū)識(shí)別河流網(wǎng)絡(luò);對(duì)DE M 單元賦流向是根據(jù)水流向最低的物

4、理原理,包括D8、Rho8、DE MON 以及D 等方法,其中最具代表性的是D8法,簡(jiǎn)單快捷,且依據(jù)水流的物理原理和水文學(xué)模型,具有重要的實(shí)用價(jià)值,本文即表述了基于該算法的一種實(shí)現(xiàn);基于局部方法和全局方法的復(fù)合方法,它是在全局徑流一致性的基礎(chǔ)上,考慮到局部地貌的特殊性而采取的一種加權(quán)方法,使提取的水系與實(shí)際河流更好地?cái)M合。但該方法具有很強(qiáng)的針對(duì)性,因此其適用性受到限制。2水文模擬算法的原理基于地表水流模擬的地形特征提取算法基本原理是2:根據(jù)地表水流從高處流向低處的自然規(guī)律,計(jì)算DE M 中每一單元的匯水量,可以認(rèn)為匯水量大于某個(gè)閾值的單元屬于匯水線,如果將這些匯水量大于給定閾值的離散點(diǎn),按規(guī)則

5、連接起來(lái)就可以將匯水線即山谷線提取出來(lái)。該原理的關(guān)鍵是3個(gè)矩陣的生成:填平原始DE M 中的洼地生成無(wú)洼地高程矩陣;在無(wú)洼地高程矩陣基礎(chǔ)上計(jì)算水流方向矩陣;在水流方向矩陣基礎(chǔ)上計(jì)算水流累積矩陣。在實(shí)際提取過(guò)程中,因在填平洼地時(shí)用到了每個(gè)單元的水流方向信息,所以應(yīng)先計(jì)算水流方向矩陣。在得到上述3個(gè)矩陣后,即可進(jìn)行地形特征或水文特征提取、匯水區(qū)域計(jì)算等分析,算法的流程圖(見(jiàn)圖1。圖1水文模擬算法流程本文采用D8算法計(jì)算單元水流方向,即根據(jù)DE M 柵格單元和八鄰域單元之間的最大坡降來(lái)確定水流方向,文獻(xiàn)3等都有詳述。本文的重點(diǎn)是洼地填平算法的設(shè)計(jì)。3模型建立與算法設(shè)計(jì)朱慶、趙杰4等曾提出在水文模擬方

6、法的基礎(chǔ)上,提取洼地的一些矢量特征,將矢量操作與柵格操作結(jié)合起來(lái)對(duì)DE M 中的洼地進(jìn)行處理。本文即是受此啟發(fā),將洼地、平地都抽象為矢量對(duì)象,并更深入討論了它們之間的關(guān)系,在此基礎(chǔ)上設(shè)計(jì)了洼地填平等操作,算法具體步驟如下:首先基于柵格操作,得到每個(gè)單元的流向,檢測(cè)出平地與洼地并提取其矢量特征;然后將平地與洼地構(gòu)建為矢量對(duì)象,操作這些對(duì)象完成洼地填平步驟;對(duì)于矢量平地對(duì)象,再將其化為柵格單元,利用矢量特征出流點(diǎn),按規(guī)則給其每個(gè)單元賦流向,最終得到水文特征分析所要的水流方向矩陣。在上述操作基礎(chǔ)上完成河網(wǎng)提取(匯水能力閾值取用平均水流累計(jì)值和流域劃分。311基本對(duì)象結(jié)構(gòu)平地、洼地的結(jié)構(gòu)如圖2:圖2平

7、地(a、洼地(b結(jié)構(gòu)圖 按照D8算法,我們將根據(jù)每個(gè)單元的流向賦值(如圖3,得到初始方向矩陣。在原始DE M 數(shù)據(jù)源中,存在許多無(wú)法給DE M 單元賦流向的情況(周?chē)鷨卧戎行膯卧呋虻雀?此時(shí)將其值賦為0,稱(chēng)這些點(diǎn)為零點(diǎn)。從這些零點(diǎn)出發(fā),采用種子算法5,可以檢測(cè)出平地與洼地,這里不再贅述。將檢測(cè)出來(lái)的平地、洼地按上圖中的結(jié)構(gòu)進(jìn)行組織、存儲(chǔ)。其中edge 為邊緣點(diǎn),fl owpoint 為出流點(diǎn),bott om 則是洼地中的零點(diǎn)集合。需要指出的是,平地、洼地的出流點(diǎn)的意義并不同,前者指平地邊緣點(diǎn)中有最大坡降的點(diǎn),后者則是洼地邊緣點(diǎn)中高程值最小的點(diǎn),且兩者可能都不唯一。對(duì)于洼地來(lái)說(shuō),要求執(zhí)行填平

8、操作后形成平地。312對(duì)象關(guān)系對(duì)于簡(jiǎn)單的洼地,將洼地中高程值小于其出流點(diǎn)高程的單元用出流點(diǎn)高程代替即可形成平地,生成平地的出流點(diǎn)應(yīng)出自原洼地的出流點(diǎn)。如果某一洼地的出流點(diǎn)同時(shí)是別的洼地或平地的邊緣點(diǎn),情況就復(fù)雜許多。比如洼地出流點(diǎn)是一平地邊緣點(diǎn),可將洼地填平并直接與此平地合并,生成新平地。問(wèn)題的關(guān)鍵在于洼地出流點(diǎn)的狀態(tài),所以在洼地、平地的檢測(cè)同時(shí)還要記錄下其出流點(diǎn)狀態(tài)。本文從簡(jiǎn)化的情況開(kāi)始,在洼地只有一個(gè)出流點(diǎn)的假設(shè)下,分別討論兩個(gè)洼地以及一個(gè)洼地與一個(gè)平地的關(guān)系。若兩個(gè)洼地共享唯一出流點(diǎn),兩個(gè)洼地可分別填平并合并:若在此點(diǎn)兩洼地存在互指關(guān)系,即在出流點(diǎn)水流方向相互指向了對(duì)方,則填平合并后的部

9、分將還在洼地內(nèi),且是新洼地bott om 部分;否則,填平合并后將形成平地,而且由于在出流點(diǎn)周?chē)赡艽嬖谂c其高程相等的點(diǎn),平地還可能在此處擴(kuò)張。需要指出的是所謂呈環(huán)狀指向的復(fù)雜洼地,可歸于相鄰?fù)莸鼗ブ傅那闆r。如圖4,設(shè)洼地D 1、D 2、D 3的出流點(diǎn)高程分別為h 1、h 2、h 3。出流點(diǎn)是洼地邊緣點(diǎn)中高程最小的點(diǎn),D 1的出流點(diǎn)指向了D 2,應(yīng)是D 1、D 2的邊緣點(diǎn),所以有h 1h 2。依次類(lèi)推,有h 1h 2h 3h 1,所以h 1=h 2=h 3,即D 1與D 2、D 1與D 3、D 2與D 3都應(yīng)是互指的關(guān)系 。若洼地出流點(diǎn)在平地邊緣,顯然洼地填平后可與平地合并,生成新平地,且將繼

10、承原平地的出流點(diǎn)。313填平算法設(shè)計(jì)與實(shí)現(xiàn)考慮復(fù)雜洼地的合并及平地的擴(kuò)張均是在洼地出流點(diǎn)發(fā)生,而新并入的洼地或平地可能帶來(lái)更多地出流點(diǎn),本文設(shè)計(jì)了如下迭代算法用于填平洼地:搜索、考察洼地每一個(gè)出流點(diǎn)狀態(tài),符合上一小節(jié)合并條件的,合并相應(yīng)的洼地或平地;對(duì)于并入的洼地,繼續(xù)按上述方法處理其出流點(diǎn);對(duì)于并入的平地,搜索、考察其邊界點(diǎn),判斷其是否為其他洼地出流點(diǎn),即是否滿(mǎn)足上節(jié)條件,作相應(yīng)處理;在洼地出流點(diǎn)處,滿(mǎn)足上節(jié)條件的,還應(yīng)作擴(kuò)散處理(按洼平地檢測(cè)時(shí)使用的種子算法。在實(shí)際操作中還有更復(fù)雜的情況,如兩洼地共用的出流點(diǎn)是平地邊界點(diǎn)等,依然可套用上述算法。將上面的步驟迭代進(jìn)行,直至洼地形成獨(dú)立的平地;

11、所有洼地都執(zhí)行此操作,即可完成水文模擬算法的填洼過(guò)程。314平地水流方向確定針對(duì)平地水流方向問(wèn)題很多研究者提出了不同的方法:Martz 和Garbrecht 的微地貌起伏算法,對(duì)平地單元高程增加不同微量直至每個(gè)單元都有一個(gè)明確的水流方向;Tribe 提出了一種匯聚模式的賦流向方法,使河流在平原區(qū)內(nèi)通過(guò)的距離最短;Jens on 的方法是從出水口出發(fā)首先確定與出口點(diǎn)相鄰的平地點(diǎn)的流向,然后處理這些點(diǎn)的鄰域點(diǎn),如此迭代直到整個(gè)平地內(nèi)所有點(diǎn)都被賦予流向。朱慶等對(duì)這些方法進(jìn)行了比較4,提出了利用平地失量模型,從提取的出流點(diǎn)開(kāi)始,由近至遠(yuǎn)隨機(jī)選擇相鄰單元進(jìn)而標(biāo)定整個(gè)平地流水方向,使其匯聚于出流點(diǎn),方便快

12、捷,還能有效消除平行和鋸齒,本文即是采用此法。4應(yīng)用結(jié)果本算法成功應(yīng)用于邙山小流域水土保持應(yīng)用系統(tǒng),利用獲得的邙山小流域DE M 數(shù)據(jù)(1131453278°1131622111°E 、341824191°341886296°N 、5m 精度,共3776行×1408列,面積95177k m 2進(jìn)行了河網(wǎng)提取和流域劃分(圖5 。圖5邙山小流域河網(wǎng)提取、流域劃分結(jié)果圖從圖5可看出河網(wǎng)提取和流域劃分結(jié)果較好的匹配了原始DE M 影像。該算法在其他小流域的適用性有待進(jìn)一步驗(yàn)證。5結(jié)論本文提出了洼地、平地矢量結(jié)構(gòu)模型,討論了它們之間關(guān)系,以面向?qū)ο蟮乃枷?/p>

13、設(shè)計(jì)了洼地填平以及平地水流方向確定的算法,與以往算法相比速度更快,也更客觀,并且在邙山小流域水土保持應(yīng)用系統(tǒng)中運(yùn)用,取得較好的成果。參考文獻(xiàn)1徐濤,胡光道1基于數(shù)字高程模型自動(dòng)提取水系的若干問(wèn)題J .地理與地理信息科學(xué),2004,20(5:111412湯國(guó)安,劉學(xué)軍,閭國(guó)年1數(shù)字高程模型及地學(xué)分析的原理與方法M .北京:科學(xué)出版社,200513李志林,朱慶1數(shù)字高程模型M .武漢:武漢大學(xué)出版社,200114朱慶,趙杰,鐘正1基于規(guī)則格網(wǎng)DE M 的地形特征提取算法J .測(cè)繪學(xué)報(bào),2004,33(1:778215周貴云,劉瑜,鄔倫1基于數(shù)字高程模型的水系提取算法J .地理學(xué)與國(guó)土研究,2000

14、,16(4:778116易紅偉,湯國(guó)安,劉詠梅1河網(wǎng)徑流節(jié)點(diǎn)及其基于DE M 的自動(dòng)提取J .水土保持學(xué)報(bào),2003,17(3:10811117任立良,劉新仁1數(shù)字高程模型信息提取與數(shù)字水文模型研究進(jìn)展J .水科學(xué)進(jìn)展,2000,11(4:4634691401測(cè)繪科學(xué)第32卷technique in visual studi o1net,takes Taihu lake watershed as a study area in order t o establish an unified,efficient and reliable sys2 te m f or data sharing1Th

15、e data sharing syste m p r ovides a great deal of data f or comp rehensive manage ment of Taihu lake watershed,and benefits f or the devel opment of the watershed1Key words:Taihu lake watershed;s mart client;data sharing; syste m i m p lementati on;FEN G L i,L I U X iao m ei,HE B aojing(Depart m ent

16、 of U rban and Res ources Sciences,Nanjing University,Nanjing 210093,China;Nanjing I nstitute of geography and li m nol ogy, CAS Nanjing210008M erg i n g ASTER w ith T M/ET M+da t a to m on itor l and cover changeAbstract:Monit oring landuse and landcover change usually u2 ses high s patial res olut

17、i on i m ages,such as I K ONOS,QU I CK B I RD or SP OT51I n this paper,taking Haidian district f or exa mp le,the au2 thors merge ASTER with T M/ET M+median res oluti on i m ages by B r ovey transfor m and p rinci pal co mponent analysis(PC Amethods, atte mp ting t o make full use of multis pectral

18、and higher s patial res olu2 ti on characteristic of ASTER and T M/ET M+i m ages,even t o m ine their potential which can be used in landuse and landcover change1 This paper may p r ovide scientific reference for cos m ically monit oring landuse and landcover change1The authors have merged multis pe

19、ctral(bands3N,2,1ASTER i m age of2003with panchr o matic ET M+i m age of1999using B r ovey transf or m Method,and fused T M543col or co mpositi on i m ages of1992with ET M+P AN of1999by PC A method,finally,as a result,both fusi on i m ages can sho w change clearly1By tested,the p recisi on of change

20、 inf or mati on detecti on ar2 rived92150%,which accords with the de mand of app licati on1I n this study,the result sho ws that cheaper ASTER and T M/ET M+data fu2 si on may efficiently monit or landcover change,es pecially in the re2 gi on without high res oluti on data,which can save the cost of

21、dyna m ic monit oring1Mean while,it indicates that ASTER and T M/ET M+i m2 ages have large utility value,and deserve extensive use1Key words:ASTER;T M/ET M+;br ovey transf or m;PCA; landcover changeHE Fenqin,HU Z henqi,Y I N J ianzhong,FAN G Sh i bo(I nstitute of Land Reclamati on and Ecol ogical Re

22、st ora2 ti on,China University ofM ining and Technol ogy,Beijing100083, China;State Key Laborat ory of I nf or mati on Engineering in Sur2 veying,M app ing and Remote Sensing,W uhan University,W u2 han430079,China;I nstitute of Botany,CAS,Beijing 100093,ChinaThe software m ethods rea li za ti on to

23、i m prov i n g rea lti m e re2 spon se i n e m bedded G I SAbstract:Due t o comp rehensive app licati on,it is very signifi2 cant t o research realti m e res ponse about e mbedded GI S1After choo2 sing an e mbedded CP U,there are s ome s oft w are methods t o i m p r ove realti m e res ponse1Map s p

24、ace data organizing structure,double buffer technique,s pecial arithmetic of drawing map sy mbols,etc are described in details in this article1Testing indicates that these ways are very effective1Key words:e mbedded GI S;realti m e res ponse;bl ocked and graded s pace data architecture;double buffer

25、 technique1HU Zem ing,YU E Chunsheng,WAN G Zhigang(I nfor ma2 ti on Engineer University,Zhengzhou in Henan Pr ovince,450002Syste m desi gn of veh i cle i n forma ti on manage m en t pl a tfor m and its key technology ana lysisAbstract:This paper is mainly about the infor mati on manage2 ment of vehi

26、cles based on GPS l ocati on infor mati on1Firstly,it gives the syste m design of vehicle inf or mati on trace and manage ment1 Then,it analyzes the relative key technol ogies1Finally,it gives the effective exa mp le and its perf or mance analysis1Key words:moving objects l ocati on;GPRS;W eb Servic

27、es; COM+;database connecti on poolHUAN G Guangyu,SHEN Zhanfeng,LUO J ianchengSHEN G Hao(School of the Earth Sciences and Res ources,Chi2 na University of Geosciences,Beijing100083,China;I nstitute of Re mote Sensing App licati ons,Chinese Acade my of Sciences,Beijing 100101,ChinaAn a lgor ith m for

28、extracti n g wa tershed fea tures ba sed on D E M Abstract:This paper p resents an algorith m f or watershed fea2 tures extracti on based on DE M1On the basis of hydr ol ogy si m ulati on algorith m,we p r opose a structural model of the dep ressi ons and flats,and study the t opol ogical relati ons

29、 when f or m ing the comp lex dep ressi ons1Then we design an algorith m t o fill dep ressi ons1This app r oach has been used in the p r oject of water and s oil p reservati on in Mangshan s mall watershed1And the result p r oves its efficiency and convenience1Key words:grid DE M;hydr ol ogy si m ul

30、ati on;t opographic fea2 ture;watershed featureL I Q inchao,L I Hongw ei,M EN G Chanyuan(I nsti2 tute of Surveying and Mapp ing,I nf or mati on Engineering University, Zhengzhou450052,China;Naval I nstitute of Hydr ographic Sur2 veying and Charting,Tianjin300061,ChinaPositi on and area prec isi on s

31、tudy of Spot5i m age correcti on i n urbanAbstract:I n order t o p r obe the positi on and area p recisi ons of the SP OT5i m age corrected with2D polynom ial in urban regi on,dif2 ferent p recisi on(acquired fr om GPS field surveying and1:10000 t opography mapand nu mber(25and6of gr ound contr ol p

32、oints (GCPswere used t o correct the SP OT5i m age with2D polynom ial functi on res pectively110GPS points were used as check points (CPst o test the positi on and area p recisi on of f our results1The conclusi ons were dra wn that:(1For urban SP OT5i m age,thr ough 2D2sd polynom ial functi on using

33、6high p recisi on GPS GCPs,better positi on p recisi on could be obtained1(2Satisfied positi on p recisi on could be reached using1:10000t opography map if GCPs are e2 nough and l ocati on of the m is well distributed1(3Satisfied area p recisi on could be reached on the i m ages corrected by GPS CPs

34、 and map CPs and that is enough for urban green s pace syste m surveying1 (4The relati onshi p bet w een positi on and area err ors is comp lex on the corrected i m ages,s o the area err ors cannot be deduced by point err ors according t o theory of err ors1Key words:RS;SP OT5;geometric correcti on;

35、point and area p recisi onFEI X ianyun,ZHANG Zhiguo,G AO X iangwei,Q I U J ianzhuang(College of I nf or mati on Science and Engineering, Shandong Agricultural University,Taian,271018;College of Re2 s ources and Envir on ment,Shandong Agricultural University,Taian, 271018;I nstitute of Land Recla mat

36、i on and Ecol ogical Reconstruc2 ti on,China University ofM ining and Technol ogy,Beijing100083The appli ca ti on of3D Zern i ke m o m en ts i n3D terra i n ma tch i n g Abstract:I n this paper the theory of matching research of3D terrain are given firstly1Then the definiti on of3D Zernike mo ments

37、is intr oduced1A3D terrain matching alg orith m based on3D Zernike mo2 ments is pr oposed1The i m p le mentati on of this alg orith m is presented1 Finally,as is revealed by our study,the alg orith m based on the3D Zernike mo ments is a mong the most successful technique at p resent1 Key words:3D te

38、rrain surface;3D Zernike mo ments;terrain matchingHU X iulin,CHE L ong,YE B in(Dep t1of E1I1,Huazhong University of Science and Technol ogy,W uhan430074,ChinaA lgor ith m s of shortest pa th for ra ster network ba sed on map a l2 gebraAbstract:The status quo of net w ork analysis and algorith m s of shortest path is discussed in this paper firstl