Survey on Local Binary Pattern Descriptors for Face Recognition
Mohamed Kas, Youssef El Merabet, Yassine Ruichek, Rochdi Messoussi · 2019
During the two past decades, the Local Binary Patterns descriptor demonstrated remarkable performance and high robustness in extracting distinguishing features from a given image. Therefore, this feature extraction method has been widely applied in diverse challenging computer vision applications including face recognition. The efficiency and usability of the LBP operator and its success in various real world applications has inspired the development of much new powerful LBP variants. Indeed, after the appearance of the LBP operator, several renowned extensions and modifications of LBP have been proposed in the literature to the point that it can be difficult to recognize their respective LBP-related strategies, strengths and weaknesses according to a given application, and there is a need for a complete comparative study in face recognition application. This paper reviews the performance of 30 recent state-of-the-art handcrafted descriptors in face recognition through a comprehensive experimental study using widely used benchmarks. Simulated experiments on ORL, Extended Yale B and FERET databases proved that some evaluated descriptors realize good classification results which outperform many recent state-of-the-art face recognition systems.