Descriptor feature based on local binary pattern for face classification

Local Binary Patterns (LBP) is a non-parametric descriptor whose purpose to effectively summarize local image configurations. It has generated increasing interest in many aspects including facial analysis, vision detection, expression demographic classification, etc. recent years and proven useful various applications. This paper presents binary pattern based face recognition technology using Vector Support Machine (SVM). Combine the characteristics of LBP with universal so that general picture are more robust. To reduce dimension maximize discrimination, super vector machines (SVM) used. Screened Evaluated (FAR), FARR Accuracy Score (Acc), not only on Yale Face database but also expanded Database B datasets, test results indicate approach accurate practical, gives rate 98 %.