Premium
A Face Identification System Based on the Mahalanobis–Taguchi System
Author(s) -
Nagao Mitsuyoshi,
Yamamoto Masahito,
Suzuki Keiji,
Ohuchi Azuma
Publication year - 2001
Publication title -
international transactions in operational research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.032
H-Index - 52
eISSN - 1475-3995
pISSN - 0969-6016
DOI - 10.1111/1475-3995.00004
Subject(s) - mahalanobis distance , taguchi methods , computer science , identification (biology) , artificial intelligence , face (sociological concept) , pattern recognition (psychology) , facial recognition system , computer vision , noise (video) , orthogonal array , data mining , machine learning , image (mathematics) , social science , botany , sociology , biology
In this paper, we propose a face identification system based on the Mahalanobis–Taguchi System (MTS). The MTS is one of the pattern recognition methods frequently used in quality engineering, and can perform robust pattern recognition by using training data, including noise. It is likely that this advantage will allow the effective implementation of a robust face identification system against lighting and face position fluctuations. Moreover, the MTS can optimize the number of attributes required for identification by using the orthogonal array and the signal/noise (SN) ratio. The face identification system has to deal with many users and the amount of data in a facial image is large. Therefore, the time required for identification can be decreased and the amount of data in the facial image database can be reduced by performing the optimization. We confirmed the effectiveness of the proposed system through practical experiments. The experimental results revealed that the MTS was an effective method for robust face identification, and could effectively reduce the number of attributes required for identification.