微处理机
微處理機
미처리궤
Microprocessors
2015年
5期
43-46
,共4页
SIFT特征%Harris特征%自适应非最大抑制%Delaunay三角形%特征提取%特征匹配
SIFT特徵%Harris特徵%自適應非最大抑製%Delaunay三角形%特徵提取%特徵匹配
SIFT특정%Harris특정%자괄응비최대억제%Delaunay삼각형%특정제취%특정필배
SIFT Feature%Harris Feature%ANMS(Adaptive Non -Maximal Suppression)%Delaunay triangle%Feature extraction%Feature matching
针对影像对匹配过程中,由于场景复杂,干扰严重,导致错误匹配率较高的问题,提出一种在三角形约束下,基于 SIFT(Scale Invariant Feature Transform)算法的图像匹配策略。该方法大致分两步进行,首先是粗匹配,使用 SIFT 算法和 Harris 选择区别度较高的特征点,得到良好特征点集。第二步为细匹配,根据良好点集建立 Delaunay 同名三角网格,对同名三角形再次使用 SIFT 算法提取特征点进行匹配。实验表明,提出的算法一定程度上提高了特征点复现率和匹配准确率。
針對影像對匹配過程中,由于場景複雜,榦擾嚴重,導緻錯誤匹配率較高的問題,提齣一種在三角形約束下,基于 SIFT(Scale Invariant Feature Transform)算法的圖像匹配策略。該方法大緻分兩步進行,首先是粗匹配,使用 SIFT 算法和 Harris 選擇區彆度較高的特徵點,得到良好特徵點集。第二步為細匹配,根據良好點集建立 Delaunay 同名三角網格,對同名三角形再次使用 SIFT 算法提取特徵點進行匹配。實驗錶明,提齣的算法一定程度上提高瞭特徵點複現率和匹配準確率。
침대영상대필배과정중,유우장경복잡,간우엄중,도치착오필배솔교고적문제,제출일충재삼각형약속하,기우 SIFT(Scale Invariant Feature Transform)산법적도상필배책략。해방법대치분량보진행,수선시조필배,사용 SIFT 산법화 Harris 선택구별도교고적특정점,득도량호특정점집。제이보위세필배,근거량호점집건립 Delaunay 동명삼각망격,대동명삼각형재차사용 SIFT 산법제취특정점진행필배。실험표명,제출적산법일정정도상제고료특정점복현솔화필배준학솔。
Aiming at decreasing the high false matching rate caused by scene complexity and serious disturbance lying in images matching,a method based on triangle constraint is proposed.It is a novel image matching strategy based on SIFT (Scale Invariant Feature Transform)algorithm and uses the information of image structure.The method is separated into two steps.The first step is rough matching and good feature point is selected by applying SIFT and Harris algorithm to acquire distinctive feature point.The second step is elaboration matching,and the corresponding Delaunay triangulation network is established based on good feature points,and then SIFT is performed again to select feature points.It was verified that the method improved the repetition and accuracy of feature matching.