进一步采用基于距离和曲线形态的双尺度相似性度量谱聚类算法进行聚类处理，通过求n 个二次规划问题，就可以求得相似度矩阵W，降低了谱聚类算法对参数的敏感性，使算法更稳定(Furthermore, we use the two-scale similarity measure spectrum clustering algorithm based on distance and curve shape to carry out clustering. By finding n quadratic programming problems, we can obtain the similarity matrix W and reduce the sensitivity of the spectral clustering algorithm to the parameters Sex, make the algorithm more stable)