Face Recognition Based on Local Binary Patterns with Threshold

This paper presents a novel and efficient face recognition technique based on Local Binary Pattern (LBP) with threshold for resolving traditional LBP´s weakness of extracting global features. By setting a threshold to enhance the robustness to noise such as light and extract the global features of face preferably. Combining the local features by LBP with global features as the total features of the image is more power discriminating. Principal Component Analysis (PCA) and linear discriminate analysis (LDA) are used to reduce the dimensionality and optimize discriminative recognition respectively. The proposed method is tested and evaluated not only on ORL datasets but also on YALE datasets and yields a recognition rate of 98%, the experimental results show that the method is valid and feasible.