Research On The Evaluation Methodology For Face Recognition Systems

Face recognition is a new hot spot of pattern recognition and artificial intelligence. Since 60s of the last century, about 40 years, these kinds of research have made remarkable achievement. Face recognition technology has got out of laboratory and applied to commercial applications successfully. A lot of algorithms and commercial systems about face recognition have come to application.As we know, face recognition is conditioned by two factors: one is the level of algorithms about face recognition technology; the other is the Methodology for integrated performance evaluation about the algorithms. Not only could the evaluation found the best one in all the algorithms and systems; but also it tested the present situation of face recognition technology and pointed out the direction of researching. The works of this article follows:(1) The algorithms of face recognition were introduced. This paper tried to summarize the algorithms about face recognition and analyses the characters of them.(2) Several tests of face recognition systems were introduced. Including EFRET, FRVT2000, FRVT2002 and FRVT2006.(3) The Methodology for Evaluation of face recognition algorithms were studied.The evaluation principle of algorithms abut face recognition, the classification of evaluation methodology and the index in different evaluation were introduced in this paper.