基于Gabor变换与自组织稀疏RAM的N-tuple神经网络的人脸识别方法  

Face Recognition Based on Gabor Transforms and The Approximate Type N-tuple Neural Network with Self-organizing Sparse RAM

作　　者：周兆捷[1] 吴乐南[1] 孙大瑞[1] 

机构地区：[1]东南大学无线电工程系,江苏南京210096

出　　处：《电路与系统学报》2004年第1期115-118,共4页Journal of Circuits and Systems

摘　　要：本文提出了一种采用首先对人脸图像进行Gabor变换,然后由自组织稀疏RAM的N-tuple神经网络进行训练识别的方法,通过大量实验证明,该方法在较少训练样本下条件下,能够取得较高的识别率。A face recognition approach based on Gabor transforms and the approximate type N-tuple neural network with self-organizing sparse RAM is presented. At first, the Gabor transforms of face image are used as its feature vectors. Then these feature vectors are input into the approximate type N-tuple neural network with self-organizing sparse RAM to classify these faces. Experimental results demonstrate that the performance of this face recognition method is good only with a few training samples.

关 键 词：人脸识别 GABOR变换 自组织映射 N-tuple分类器 稀疏分布存储器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]