Abstract: In real scenes, unknown number of categories, unbalanced number of samples and complex data distribution can lead to low accuracy of intelligent face image extraction. To solve these problems, a face image clustering algorithm based on the discovery of key edges of association graph is proposed. First, the algorithm merges features of image samples extracted by multiple convolution neural networks to obtain new feature vectors with stronger capability of identification, and then calculates the similarity between different samples. Then, an appropriate threshold value is set for the rejection rate and the recognition rate, and the similarity result obtained in the previous stage is compared with the threshold value. The sample pairs with high similarity degree are screened out, and the association graph is constructed by adding the connecting edge between the above sample pairs. Using the intermediate number centrality measure, the key edge discovery method is designed to dig the important connecting edges that may connect different clusters in the association graph. Finally, the graph convolution network is used to confirm the existence of the above important connection edges, such that the final cluster is obtained. Experimental results show that the proposed algorithm can improve the accuracy of face image
clustering.

Key words: face clustering, key links discovery, betweenness centrality, graph convolutional network