Estimation of image quality factors for face recognition.

摘要

Over the past few years, verification and identification of humans using biometric has gained attention of researchers and of the public in general. Face recognition systems are used by the public and the government and are applied in different facets of life including security, identification of criminals and identification of terrorists. Because of the importance of these applications, it is of great necessity that face recognition systems be as accurate as possible. Some research has shown that image quality degrades the performance of face recognition systems. Most previous research has focused on designing algorithms for face recognition that deal or compensate a single effect such as blur, lighting conditions, pose, and emotions. In this thesis we identify a number of factors influencing recognition performance and conduct an extensive study of the effects of image quality factors on recognition performance and discuss methods to estimate this quality factors.