Face Recognition Research Based on the Fusion of Layered LBP Feature

Local Binary Pattern(LBP) is an operator to describe the local texture of face, which has the characteristics of rotation invariance and grayscale invariance. This paper proposed a face recognition arithmetic based on the fusion of layered LBP feature(F-LLBP), using the LBP operator to extract the layered texture feature, fusing the characteristics of each layer, reducing the dimension with Principal Component Analysis and Linear Discriminant Analysis(PCA-LDA) to get a set of low dimensional characteristic which are easy to classify, finally using the KNN arithmetic to classify. The experiments on ORL face database verifies the effectiveness of the F-LLBP, the experimental results indicate that the recognition rate of FLLBP increased 8% relative to the traditional LBP feature combined with PCA dimension reduction arithmetic.