ABSTRACT
Face recognition is a biometric authentication technology based on human facial features. Compared with other recognition technologies, human face recognition has its natural and imperceptible advantages. The past two decades have seen the wide application of this technology and its more and more social and commercial value. Most existing face recognition methods succeed on the premise of massive training samples within its system. However, the system can only maintain one face image a person in actual practice, due to the diculty in sample collection and limitation of other requirements, which is then also called single-sample face recognition [1]. As to common face recognition methods, the limited information in single-sample recognition leads to the weakening performance or even failure in recognition. Therefore, the accurate recognition of face in single sample becomes a challenging problem.
