Local detail features of face are important bases for recognizing different persons. For its invariant to monotonic gray-scale transformations and it’s a non-parametric kernel which summarizes the local special structure of an image, the Local Binary Pattern (LBP) has becoming a popular technique for face representation. In this paper, the LBP technique and its application for representing faces are investigated. Two experiments using the histogram sequence of block LBP are performed on ORL face database to validate the effectiveness. In experiment 1, Carle square dissimilarity measure is used as a classifier directly, to classification the histogram sequences of LBP, which have been extracted in partitioned face images.
In experiment 2, Principal Component Analysis (PCA) method is used to classification the histogram sequences, which have been converted to vectors. The results show that the number of blocks partitioned in face image is related to the recognition rate. Too much and too little are not beneficial to recognition. In addition, two typical methods are performed on ORL face database to compare the performance of LBP method with others. And the experimental results show that method used in experiment 2 obtains a better classification performance compared with experiment 1 and two typical methods.