資源描述:
《彩色地形圖點(diǎn)狀地物符號(hào)的提取與識(shí)別》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、彩色地形圖點(diǎn)狀地物簦等的提取與識(shí)別中文摘要彩色地形圖點(diǎn)狀地物符號(hào)的提取與識(shí)別中文摘要地理空間數(shù)據(jù)對(duì)數(shù)字城市的建設(shè)至關(guān)重要,將現(xiàn)有的地形圖數(shù)字化,并對(duì)地圖要素進(jìn)行自動(dòng)提取和識(shí)別以獲得地理信息,是獲得地理空間數(shù)據(jù)的一種低廉而高效的方法,在今后相當(dāng)長(zhǎng)的一段時(shí)間內(nèi)仍將是數(shù)字地理信息的主要來源之一。本文針對(duì)彩色地形圖點(diǎn)狀地物符號(hào)的自動(dòng)提取和識(shí)別作了具體的研究工作。本文首先分析了人閱讀地圖的過程,并利用模式識(shí)別理論分析了計(jì)算機(jī)閱讀地圖過程,在此基礎(chǔ)上對(duì)地圖識(shí)別的總體框架進(jìn)行了設(shè)計(jì),采用二次去嗓模型,提高了彩色圖像處理的效果。地形圖經(jīng)掃描后,采用非線性各向異性擴(kuò)散方法對(duì)彩色圖像進(jìn)行平滑處理,以
2、提高彩色圖像的質(zhì)量。通過對(duì)各種顏色空間的比較,選擇了適合彩色地形圖分割的11.12.13顏色空聞,并利用模糊C均值聚類方法在顏色鏈表的直方圈上完成了對(duì)彩色地形圖的自動(dòng)分色,得到了黑、棕、藍(lán)、綠四個(gè)要素圖。要素圖經(jīng)二值化后,用數(shù)學(xué)形態(tài)學(xué)方法對(duì)其進(jìn)行二次去噪,修補(bǔ)斷點(diǎn),去除孤立點(diǎn)、毛刺,為后續(xù)的識(shí)別打下基礎(chǔ)。對(duì)細(xì)化的二值要素圖進(jìn)行矢量化操作,得到矢量化數(shù)據(jù)。最后使用統(tǒng)計(jì)方法和神經(jīng)網(wǎng)絡(luò)方法實(shí)現(xiàn)了點(diǎn)狀她物符號(hào)的識(shí)別。關(guān)鍵詞:地形圖自動(dòng)分色地圖模式識(shí)別點(diǎn)狀符號(hào)數(shù)字圖像處理作者:周蓓蓓指導(dǎo)教師:龔聲蓉英文摘要彩色地形圖點(diǎn)狀地物符口.的提取與識(shí)別ResearchonAutomaticExtra
3、ctionandRecognitionofColorTopographicMapPointSymb01.ABSTRACTGeographicspecialdataisacrucialpartinCyberCityconstruction.Convertingtheoldmapstodigitalformatandextractingmaps’essentials,resultinginachievinggeographicinformation,istheoneoftheeconomicandeffectiveways.Anditwillbeamainsourceofdigita
4、lgeographicinformationforalongwhilefromnowon.Thisthesisresearchedtheautomaticextractionandrecognitiontopographicpointsymbolindetail.Atfirst,itanalyzedtheprocessofhumanreadingmaps,andthenusingmoderecognitiontheory,italsoanalyzedtheprocessofcomputerreadingmaps.Basedonabovetwoefforts,thethesisde
5、signedthewholeframeofmaprecognition.Inordertoimprovethecolormapprocessresult,itemployed2-stepnoiseimmunitymodel.Aftermapscanning,theimageefficiencyCanbeimprovedbysmoothingthecolorbythemeansofno-linearwayanddifferentcharactersindifferentdirection.Comparisonofvariouscolorspacearemade.ThusI1一12-
6、13colorspaceischosenastheappropriateonesuitableforcolorspacesegmentation.ThefussyCmeansclusteringautomaticallysegmentthecolortopographicmaps,generatingfourcoloressentialelementssuchasblack,blown,blueandgreen.Formattingtheseessentialelementsintotwovalues(0o.r1),mathematicmorphologyisimplemente
7、dtodecreasenoise,whichbenefitthefollowingimagerecognition.Theprocessedelementsarevectorizedafterthinning.Thestatisticmethodandartificialneuralnetworksareemployedtorecognizethepointtopographicsymbolinblackelementmap.Keywords:TopographicMap;Aut