摘要
本文提出一种基于Delaunay三角剖分和摄影几何中的射影不变量的一种相机图像匹配算法。本算法首先利用三角形相似函数计算待匹配图像的Delaunay三角形网中三角形之间的相似度,进行粗匹配,然后利用射影不变量删除粗匹配结果中的误匹配,进行精匹配。对实际图像的实验表明本文的算法具有良好的匹配性能,可以对部分物体运动的图像进行匹配。
This paper proposed a new image matching method based on Delaunay triangulation and projective invariant in projective geometry .This new method firstly used triangulation function to compute the triangulation similarity of the triangulation in the matching images Delaunay triangulation ,then used the triangulation to execute rough matching. Secondly used projective invariant to remove the error matching in the result of rough matching And we realize this matching algorithm could match the photo images with part object moved. Experiments on real images demonstrate the good performance of this method.
出处
《微计算机信息》
2009年第28期115-117,共3页
Control & Automation