Face recognition based on the appearance of local regions

Abstract

Recently, we proposed a novel facial representation for face recognition based on the local binary pattern (LBP) features. We obtained excellent results when dividing the face images into several regions from which the LBP features are extracted and concatenated into an enhanced feature vector as a face descriptor. However, it was unclear whether the obtained results were due to the use of local regions (instead of a holistic approach) or to the discriminative power of LBP. In this work, we investigated this issue by adopting and comparing four different texture features when using the appearances of local regions. The experimental results clearly showed and confirmed the validity of using LBP for face description.