谱图理论已经广泛的应用于形状表示、匹配和聚类。主要研究两个方面的问题:1、通过提取形状的骨架特征点和图谱分析,建立形状的数学模型;2、通过分析模型的数学特征,将形状投影到低维空间中。利用上述方法实现了形状的聚类仿真。针对公共数据集的仿真实验证明了该聚类算法能有效的实现形状聚类。 Spectrogram theory has been widely used in shape representation, matching and clustering. Mainly studies two aspects of the problem: 1, through the extraction of the shape of the skeleton feature points and the analysis of the map to establish the shape of the mathematical model; 2, by analyzing the mathematical characteristics of the model, the shape projected to low-dimensional space. Using the above method to achieve the shape of the cluster simulation. Simulation experiments on public data sets show that the clustering algorithm can effectively achieve the shape clustering.