Human Face Recognition for Security System

カテゴリ: 部門大会

論文No: TC1-2

グループ名: 【C】平成22年電気学会電子・情報・システム部門大会講演論文集

発行日: 2010/09/02

タイトル(英語): Human Face Recognition for Security System

著者名: I Gede Pasek Suta Wijaya (熊本大学),Keiichi Uchimura(熊本大学),Go Koutaki(熊本大学)

著者名(英語): Gede Wijaya(Kumamoto University),Keiichihi KeiicUchimura(Kumamoto University),Keiichi Uchimura(Kumamoto University)

キーワード: sub-space|holistic features|LDA|real-time| face recognitionsecurity system

要約(日本語): This paper presents the real-time human face recognition using the predictive LDA for security system. It can be attained because the compact features of face image consisting of dominant frequency contents and statistical moment of face image is employed as input of PDLDA system. In addition, the compact features have small size (<1% coefficients of original size). In other side, the PDLDA has better capability to classify the compact features than that of the any variations of LDA and PCA. The main aim of the proposed system is to overcome some weaknesses of current security system: transferable and predictable. From 50 subjects that were tested, the FRR is about 2% and the FAR is 4% with requiring short time processing for retraining and querying.

PDFファイルサイズ: 3,967 Kバイト