仿射变换下基于凸包和多尺度积分特征的形状匹配方法

摘要

针对在仿射变换下的形状匹配问题,提出基于凸包的特征点提取方法、基于各向异性高斯核的多尺度积分特征和基于两者的匹配方法.首先提取形状的凸包,根据最大面积原则对凸包相邻顶点之间的曲线进行演化,获取的点和凸包顶点形成仿射不变的特征点;其次对特征点按顺序编组,根据特征点之间的仿射变换关系构造多尺度积分特征向量;最后使用动态规划算法计算形状之间的相似度.实验结果表明,该方法对局部形变和噪声敏感度小,并适用于复杂形状的匹配.此外,特征点提取方法和多尺度积分特征也可与其他方法结合进行形状分析.%Feature point extraction method based on convex hull,multiscale feature based on anisotropic Gaussian kernel and matching method based on them are proposed to solve the shape matching problem under affine transformation.Firstly,the convex hull of the shape is extracted.The curve segments between the adjacent vertices of the convex hull are evolved by maximizing the area of the triangle formed by the adjacent vertices and the points of the segment.The affine invariant features consist of the vertices of the convex hull and the points obtained by the evolution.Secondly,the feature points are grouped in order and the multiscale integral feature vectors are constructed according to the affine relationship between them.Finally,the dynamic programming is used to measure the similarity of the shapes.Experiments show that our method is not sensitive to the local noises and deformations and is suitable for the matching of complicate shapes.Moreover,the feature point extraction method and the multiscale feature can also be combined with other methods to analysis of shapes.