基于图的Laplace谱的特征匹配

Feature Matching Based on Laplacian Spectra of Graphs

为了能对等距变换和相似变换后的图像进行精确匹配,提出了一种基于图的Laplace谱的特征匹配方法,该方法是首先给定两幅图像的特征点,然后分别定义其Laplace矩阵,再通过分析该矩阵的特征值及特征向量来构造特征点匹配矩阵;最后根据匹配矩阵元素的大小和位置信息来实现特征点匹配,并从理论上证明了该算法在对图像进行等距变换或相似变换情况下能获得精确匹配。实验结果表明,该方法对真实图像的匹配精度可达到82%。

This paper presents a method of feature matching based on Laplacian spectral of graphs. Given feature points of two images, we define Laplacian matrices respectively, analysis the eigenvalues and eigcnvectors of the matrices, and construct a feature points matching matrix with information of magnitude and position of entries in the matching matrix, the feature points matching is done. Furthermore, we theoretically prove that our algorithm can acquire an exact matching under an equilong transformation or equiform transformation on images. Experimental results show that the approach attains accuracy 82% on real images.