Face Recognition with Laplacian Eigenmaps on Local Binary Pattern

Local Binary Pattern (LBP) operator is very simple to calculate and implement. It can efficiently extract facial texture feature which represents the local structure of face images. Laplacian Eigenmaps (LE) is a classical non-linear data dimensionality reduction method. Its main optimization do not involves local minima. Benefiting from the advantages of both them, a new approach to face recognition is constructed by combining LBP operator and LE. At first, the uniform LBP operator is used to extract the facial texture feature; then LE algorithm is used for data dimensionality reduction; finally, support vector machine (SVM) is used for classification. Extensive experiments are carried out by choosing the former 3,5,7,9 images of each subject as training set. Compared with other algorithms, the results show that the combination of LBP_LE provides better performance than that of them and prove the effectiveness of the presented algorithm.