欢迎来到天天文库
浏览记录
ID:37390059
大小:2.66 MB
页数:111页
时间:2019-05-23
《基于多幅图像的空间点定位与曲线结构三维重建》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、南京航空航天大学博士学位论文基于多幅图像的空间点定位与曲线结构三维重建姓名:张维忠申请学位级别:博士专业:航空宇航制造工程指导教师:张丽艳20070101基于多幅图像的空间点定位与曲线结构三维重建误差。实验证明,这是一种行之有效的三维测量与模型重建实用方法。(5)在深入进行理论研究的基础上,建立了一个面向工业产品、可现场操作的三维摄影测量原型系统。实例验证和分析结果表明,本系统可以取得较为满意的三维重建结果。本文方法具有测量硬件简单、测量方式灵活、测量范围不受限制、各角度测量数据自动拼合等优势,在工业产品逆向设计、产品质量检测等领域有广阔的应用前景。关
2、键词:三维测量,三维重建,根据运动的结构重建,摄影测量,标记点,同名点匹配,活动轮廓模型,三角测量法ii南京航空航天大学博士学位论文AbstractInrecentyears,withdigitalcamerabeingusedwidely,itsresolutionandimagingqualityareenhancedgreatly.Thepriceisbecominglowerandlower.Inaddition,theprogressinimageprocessingandpatternrecognitionisincreasinglyfas
3、t.Therapid,flexibleandlowcost3Ddataacquisitiontechniquesbyusingdigitalcamerashavebecomethefurthestpromising3Dmeasurementways.Inthisthesis,3Dpointandcurvestructurereconstructionfrommultipleimagesobtainedbyonedigitalcameraisdeeplystudiedbasedontherigorousphotogrammetrytheoriesandt
4、helatestresearchachievementsincomputervision.Anewmethodofflexible3Dmeasurementisproposedforindustrialproducts.Toensurereliabilityoffeaturerecognitionandtoheightentheprecisionofimageprocessingandthelevelofmeasurementautomatization,asetofreferencepoints(includingcodedpointsandunco
5、dedpoints)andmarkedobjectcurvesareapplied.Thecodedpointsareusedasanassistantutility,whereastheuncodedpointsandthemarkedcurvestructuresaretheelementstobereconstructed.Themaincontentsandcontributionsofthisthesisareasfollows:(1)Analgorithmofautomaticreferencepointdetectionisputforw
6、ardandimplemented.Firstly,thetargetsareextractedfromimagesaccordingtotheirsize,shapeandintensityetc.Thenanimprovedmethodtoidentifytheuniqueidentityofeverycodedtargetisputforward.Atlast,themultiple-threshold-linealgorithmisappliedtogetthesubpixellocationofbothcodedanduncodedtarge
7、ts.Practicalexamplesshowthattheresultsaresatisfying.(2)Anon-iterativeimagefeaturematchingalgorithmbasedonreferencepointcorrespondencesisproposed.Theinitialmatchingoftheuncodedtargetcorrespondencesbetweenanimagepairisestablishedaccordingtothesimilarityandcompatibility,whicharebas
8、edontheIDcorrespondencesofthecodedtargets.Theou
此文档下载收益归作者所有