资源描述:
《matlab图像拼接算法及实现》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、图像拼接算法及实现(一)来源:中国论文下载中心 [09-06-0316:36:00] 作者:陈挺 编辑:studa090420 论文关键词:图像拼接 图像配准 图像融合 全景图 论文摘要:图像拼接(imagemosaic)技术是将一组相互间重叠部分的图像序列进行空间匹配对准,经重采样合成后形成一幅包含各图像序列信息的宽视角场景的、完整的、高清晰的新图像的技术。图像拼接在摄影测量学、计算机视觉、遥感图像处理、医学图像分析、计算机图形学等领域有着广泛的应用价值。一般来说,图像拼接的过程由图像获取,图像配准,图像合成三步骤组成,其中图像配准是整个图像拼接的基础。本文研究了
2、两种图像配准算法:基于特征和基于变换域的图像配准算法。在基于特征的配准算法的基础上,提出一种稳健的基于特征点的配准算法。首先改进Harris角点检测算法,有效提高所提取特征点的速度和精度。然后利用相似测度NCC(normalizedcrosscorrelation——归一化互相关),通过用双向最大相关系数匹配的方法提取出初始特征点对,用随机采样法RANSAC(RandomSampleConsensus)剔除伪特征点对,实现特征点对的精确匹配。最后用正确的特征点匹配对实现图像的配准。本文提出的算法适应性较强,在重复性纹理、旋转角度比较大等较难自动匹配场合下仍可以准确实现图像配准。 Abs
3、tract:Imagemosaicisatechnologythatcarriesonthespatialmatchingtoaseriesofimagewhichareoverlappedwitheachother,andfinallybuildsaseamlessandhighqualityimagewhichhashighresolutionandbigeyeshot.Imagemosaichaswidelyapplicationsinthefieldsofphotogrammetry,computervision,remotesensingimageprocessing,medi
4、calimageanalysis,computergraphicandsoon.。Ingeneral,theprocessofimagemosaicbytheimageacquisition,imageregistration,imagesynthesisofthreesteps,oneofimageregistrationarethebasisoftheentireimagemosaic.Inthispaper,twoimageregistrationalgorithm:Basedonthecharacteristicsandtransformdomain-basedimageregi
5、strationalgorithm.Infeature-basedregistrationalgorithmbasedonarobustfeature-basedregistrationalgorithmpoints.Firstofall,toimprovetheHarriscornerdetectionalgorithm,effectivelyimprovetheextractionoffeaturepointsofthespeedandaccuracy.AndtheuseofasimilarmeasureofNCC(normalizedcrosscorrelation-Normali
6、zedcross-correlation),throughthelargestcorrelationcoefficientwithtwo-waymatchingtoextractthefeaturepointsouttheinitialright,usingrandomsamplingmethodRANSAC(RandomSampleConsensus)excludingpseudo-featurepointsright,featurepointsontheimplementationoftheexactmatch.Finallywiththecorrectfeaturepointmat
7、chingforimageregistrationimplementation.Inthispaper,thealgorithmadapted,intherepetitivetexture,suchasrelativelylargerotationmoredifficulttoautomaticallymatchoccasionscanstillachieveanaccurateimageregistration. Keyword