欢迎来到天天文库
浏览记录
ID:23000908
大小:933.00 KB
页数:40页
时间:2018-11-02
《[计算机软件及应用]数字图像中susan特征点检测算法实现》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库。
1、摘要在图像理解、图像匹配、三维重建及模式识别等领域中,特征点的检测具有十分重要的意义。特征点在保留图像中物体的重要特征信息的同时有效地减少了信息的数据量,使得对图像处理时运算量大大减少。特征点的的定义有多种不同的表述,如图像中灰度值和像素剧烈变化的点、图像边界上具有较高曲率的点等。对于特征点的定义决定了特征点的特性,同时也决定了所检测出的特征点的检测所采用的方法。SUSAN(SmallestUnivalueSegmentAssimilatingNucleus)算法是上世纪九十年代中期提出的一种边角点检测算法,该算法是一种基于图像局部灰度特征,利用一个圆形的模板对图像进行扫描,比较模板内部
2、的点与模板中心点的灰度值,如果灰度差值小于一定的阈值,就认为该点与中心点的灰度相同,否则就认为该点与中心点有明显的差异。统计模板内部与中心点灰度相同的点的个数,与一个全局阈值进行比较,判断该点是否属于角点,从而实现对图像特征点的检测。本文采用matlab语言实现了SUSAN特征点监测并利用模拟图像和真实图像对算法进行了验证,实验表明SUSAN特征点检测算法能有效提取图像中的特征点。与传统的角点检测算法相比,SUSAN算法是一种基于图像灰度比较的算法,不涉及梯度的计算。具有速度快、抗干扰能力强等特点。关键字:SUSAN算法边缘检测角点检测matlab36AbstractInimageund
3、erstanding,imagematching,reconstruction,andpatternrecognition,thefeaturepointdetectionisofgreatsignificance.Featurepointsofobjectswhileretainingtheimportantfeaturesoftheimageinformation,whileeffectivelyreducingtheamountofdatainformation,makingtheimageprocessingoperationsisgreatlyreduced.Featurepo
4、ints,therearedifferentdefinitionsofexpression,iftheimagehasagrayvalueanddramaticchangesinthepointimagepixelontheboundaryofhighcurvaturepoints.Forthedefinitionofthefeaturepointsofthefeaturepointsofthecharacteristicsofthedecision,butalsodeterminesthefeaturepointsdetectedinthedetectionmethodused.SUS
5、AN(SmallestUnivalueSegmentAssimilatingNucleus)algorithmisthemidninetiesoflastcenturyacornerdetectionalgorithm,whichisbasedonlocalgraylevelimagefeatures,usingacirculartemplatetoscantheimage,comparetemplatewithinthetemplatecenterpointandthegrayvalue,ifthegrayleveldifferenceislessthanacertainthresho
6、ld,tothatpointandthecenterofthegraylevelofthesame,otherwisethepointthattherearesignificantdifferenceswiththecenter.Statisticscentergrayinsideandthesametemplatethenumberofdots,withaglobalthresholdvalueforcomparisontodeterminewhetherthecornerpoint,enablingthedetectionofimagefeatures.Inthispaper,mat
7、lablanguageoftheSUSANfeaturepointdetectionandthesimulatedimagesandrealimagesvalidatethealgorithm,experimentsshowthatSUSANfeaturepointdetectionalgorithmcanextracttheimagefeaturepoints.Withthetraditionalcornerdetectional
此文档下载收益归作者所有