Abstract
A human face is one of biometric features, which carries their owner's personal information. People usually don't hide their faces on their daily life and it's easy to catch their features as images sensing devices without any contacts. These characteristics are preferred by wide-range applications, i.e. smooth man-machine communication, user-friendly security improvement, and so on. The art of human face identification technologies is only available under very limited circumstances and irregular face images easily make identification performance go down, because of the generally similar shapes of faces combined with the numerous variations between images of the same face. This paper categorizes origins of variations, and some results according to computerizable approaches against them are described.
