基于非负矩阵分解和模糊C均值的图像聚类方法

One method based on non-negative matrix factorization and fuzzy C means for image clustering

非负矩阵分解(NMF)作为一种新的矩阵分解和特征提取方法,是大数据处理和模式识别中线性分离数据从而聚类的有效方法。提出了一种新的聚类算法FCM-NMF,采用NMF分解提取样本的本质特征,并用模糊C均值(FCM)进行模糊聚类。该算法将NMF目标函数与FCM算法融合,提出了新的目标函数的形式,并生成新的交替迭代公式。最后在两个标准图像数据集GHIM-10k和COREL-10k上与传统的5种聚类方法从三个评价指标进行了对比。实验结果表明,该算法在标准数据集上聚类准确率和标准化互信息值分别达到了84%和77. 21%,达到了预期目标,提高了聚类效果。

As a new method of matrix factorization and feature extraction,non-negative matrix factorization( NMF) is an effective method to cluster data by linear separation in big data processing and pattern recognition. In this paper,a new clustering algorithm FCM-NMF is proposed.The essential features of samples are extracted by NMF decomposition,and fuzzy clustering is performed with fuzzy c-mean( FCM). This algorithm integrates NMF objective function with FCM algorithm,proposes a new form of objective function,and generates a new alternating iterative formula. On two standard image data sets,GHIM-10 k and COREL-10 k,the proposed algorithm is compared with the traditional five clustering methods from three evaluation indexes. Experimental results show that the proposed algorithm achieves 84% and 77. 21% in clustering accuracy and standardized mutual information on the standard data set,and achieves the expected goal and improves the clustering effect.