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”微略来重新恢复极值特征位置的基础。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。该方法对其他一些应用领域也行之有效。

In this paper， a method about Quasar spectrum recognition is proposed. Firstly, on the signal with extreme points as its features, we introduce two important properties of multiscale morphological dilation (erosion). One is the monotone rate which is the basis of an adaptive scalar selection mechanism; the other is the monotone property which can be used to localize the features accurately by “Coarse to Fine”strategy. According to these properties, the spectrum is filtered by multiscale dilation and its features are recovered. Secondly， elastic matching is proposed to featur ever if ication and its physical meaning is pointed out. In fact, this method can be applied to other fields effectively.