Efficient Human Face Recognition Method under Subtle SIFT Features Using Optimized K-means
Zhikai Zong · International Journal of Signal Processing Image Processing and Pattern Recognition · 2017
K-means proposed by MacQueen has been successfully and widely applied to pattern recognition and machine learning.However, the performance of k-means in human face recognition has barely been surveyed systematically.In this paper, we combine optimized k-means clustering algorithm with scale invariant feature transform (SIFT) features to improve face recognition rate.To extract SIFT features from test face image, subtle human face features will be obtained after clustering of SIFT features by optimized kmeans.Human face is identified by calculating the distance between subtle features.The large scale of experiments on JAFFE and FERET face databases have been carried out to prove the effectiveness of the proposed algorithm, and compared with other methods such as k-means, SIFT features, linear discriminant analysis (LDA) and principal components analysis (PCA).The experimental results demonstrate that the high recognition rates can be obtained by the proposed method.