多尺度形态滤波及弹性匹配技术在类星体谱线识别中的应用

A Graph Based Algorithm for Generating the Delaunay Triangulation of a Planar Point Set

本文提出了一种类星体谱线证认方法。首先针对特征为极值点的信号，研究了多尺度膨胀（腐蚀）关于极值点数的两种重要特性及其应用。其一是单调率特性，根据它自动选择滤波器尺度，有效地滤除脉冲噪声；另一种是单调性，它是＂从粗到精＂策略来重新恢复极值特征位置的理论基础。根据这些性质，对光谱进行多尺度膨胀（腐蚀）和特征恢复，以滤除脉冲噪声而不影响谱线特征。然后研究弹性匹配技术应用于谱线证认，并指出了匹配方法中参量的物理意义。

A graph based algorithm for generating the Delaunay triangulation of a planar point set is presented in this paper. The first step is to calculate the Euclidean Minimum Spanning Tree (EMST) of the given points. The EMST is then augumented to triangle mesh. Finally the Delaunay Triangulation of the given points is obtained by local tranformation acording to Max Min angle criteria. The concepts and properties of Voronoi diagram of a planar point set are slso discussed in the paper.