A Cloud Face Recognition System using A New Optimal Local Binary Pattern

Local Binary Pattern (LBP) is a non-parametric descriptor largely used in different pattern recognition fields. In last years, to solve basic LBP descriptor shortcomings, many LBP varieties are proposed. Here, we propose an adaptive local feature descriptor for face recognition based on optimization criteria. To further accentuate the significant face components, the new value of current pixel is computed using optimal contributions of its eight neighboring pixels by minimization under the constraints of an evaluation function based on the entropy. In addition, the Two Dimensional Linear Discriminant Analysis (2DLDA) will be used to reduce face features vector. Then, Sequential Minimal Optimization (SMO) machine learning algorithm will be used to classify the features dataset. The experiments on ORL and Yale face databases show that the proposed LBP variety have good and robust recognition performance. Moreover, our proposed approach can be generalized for huge faces' databases using cloud computing benefits.