Face Recognition using Stringface
Chen, Weiping · Griffith Research Online · 2012
Automatically recognizing human faces has attracted a lot of attention in the academic, commercial, and industrial communities during the last few decades due to its law intrusiveness and less cooperativeness. Face recognition technology has a variety of potential applications in information security, law enforcement, surveillance, smart cards, and access control. Despite significant advances in face recognition technology, it has yet to be put to wide use in industrial or commercial communities, mainly because of high error rates in real scenarios. Existing face recognition systems have achieved promising recognition accuracy under controlled condition. However, these systems are highly sensitive to environmental factors due to changing appearance of human face, such as variations in expression, illumination, pose, partial occlusion, and time gap between training and testing data capture. A practical face recognition system should be more robust against these varying conditions. Especially in some applications such as access control to sensitive areas, monitoring border crossing, and identifying criminals or terrorists, the system should be capable of identifying individuals who use disguise accessories to hide one’s identity to remain elusive from low enforcement. Furthermore, many reported face recognition techniques rely heavily on the size and representative of training set, and most of them will suffer serious performance drop or even fail to work if only one training sample per person is available to the system. Hence, face recognition from one sample per person is an important but challenging problem both in theory and for real-world applications. Fewer samples per person mean less laborious effort for collecting them, lower costs for storing and processing them.