基于生物视觉模型的人脸识别设计  

Design of Face Recognition Based on Biologic Visual Model

作　　者：李芹[1] 练秋生[1] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004

出　　处：《电视技术》2008年第2期81-83,86,共4页Video Engineering

基　　金：河北省教育厅自然科学项目(2004124)

摘　　要：针对人脸特性,对标准模型的S2层模版提取方法和匹配算法进行改进,并重新定义了C2层特征的平移不变范围,分类器采用简单的多类线性支持向量机(SVM)。对ORL和Yale标准人脸库的分类识别结果表明,改进的模型能高效快速地进行人脸识别,识别率分别达到99.86%和97.23%。In allusion to the characters of human face, the extracting method and matching algorithm in S2 layer template of standard model are improved. And the shift-invariance range of characters in C2 layer is redefined. The simple all-pairs linear Support Vector Machine （SVM） is adopted in classifier. The results of classification and recognition based on normal face database ORL and Yale show that the improved model can recognize the face image more exactly and quickly, and the recognition accuracy could respectively achieve 99.86% and 97.23%.

关 键 词：视觉皮层 标准模型 人脸识别 支持向量机 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]