基于棱镜的单目立体视觉匹配算法-研究

基于棱镜的单目立体视觉匹配算法-研究

ID:33150231

大小:20.57 MB

页数:72页

时间:2019-02-21

基于棱镜的单目立体视觉匹配算法-研究_第1页
基于棱镜的单目立体视觉匹配算法-研究_第2页
基于棱镜的单目立体视觉匹配算法-研究_第3页
基于棱镜的单目立体视觉匹配算法-研究_第4页
基于棱镜的单目立体视觉匹配算法-研究_第5页
资源描述:

《基于棱镜的单目立体视觉匹配算法-研究》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库

1、万方数据东北大学硕士论文AbstractResearchofMonocularStereovisionmatchingAlgorithmBasedonPrismAbstractAtpresent,stereovisionisoneofthehottestresearchtopicsincomputervision.StereovisioniSadisciplinethatstudieshowtousemachinetoimitatehumanvision.andthentounderstandandrecognizethethree—dimensionalspac

2、e.ThetheoryofvisualcomputingwasproposedbyMarrinthe80’SintwentiethCentury.Thisbasictheorymakesstereovisionbecomeahotresearchincomputervision.Binocularstereovisionisaveryimportantbranchofcomputervision.However,inthepracticalapplication,traditionalbinocularstereovision’Slargesizeandhighc

3、ostleadtomonocularstereovisionbasedonprism.Comparedwiththetraditionalbinocularstereovisionsystem,thesinglecamerabasedontheprismhastheadvantagesofsmallersizeandlowercost.Stereomatchinghasalwaysbeenthemostimporlantinstereovisionandthemostdifficultprobleminstereomatching,anddirectlydeter

4、minesthequalityofthe3Dreconstruction.Afterdecadesofdevelopment,stereomatchinghasbeenstudiedbymanyresearchers.Thestereomatchingisthedepthinformationofthecorrespondingpointsintwoormoreimagestakeninthesanlespacescene.Becausethetraditionalstereomatchingismainlyforthebinocularstereovision,

5、comparedwiththetraditionalstereomatching,prismintroducedresultsthatstereomatchinghaschangedalittleintheprismMonocularStereoVisionSystem.Aimingattheproblemofstereomatchinginthesinglecamerastereovisionsystembasedonprism,thestereomatchingisimproved.Thelnainworkofthispaperis:Firstly,thege

6、ometryofthesinglecamerastereovisionisproposed,andtherelationshipbetweentheleftandrightimagesinMonocularStereovisionsystemispresented,therelationshipmodelisthefundamentalmatrix.Then,thefundamentalmatrixiscalculatedfromtheimagecoordinatesofthefeaturepointswhicharefewandaccurate.Secondly

7、,thefeaturepointsofthetwopansoftheimageareextractedbySIFTalgorithm,andthefeaturepointsarematchedwiththeEuclideandistance.Inordertoeliminatetheerrormatching,theSIFTfi:aturesoftheinitialextractionareselectedbyRANSAC,andthematchingpointsarefoundasfaraspossible.Thirdly,usingthesecondparto

8、fthec

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。