Research On Face Image Clustering Based On Integrating Som And Spectral Clustering Algorithm

Abstract

With the rapid development of neural network and face recognition, the Convolution Neural Networks(CNN) is more and more widely applied. Considering the shortcomings of traditional clustering methods, spectral clustering and self-organizing feature mapping neural network (SOM), a variety of new clustering methods based on deep learning are constantly created. In this paper, firstly a convolutional neural network is used in automatically feature extraction. After extracting features with CNN, S-SOM, an integrated algorithm of SOM and spectral clustering, is proposed to cluster face image data. Finally, to verify the effectiveness of the method, an experiment is employed. Meanwhile, we also analyze the effect of some related parameters of S-SOM on face clustering for further research.