基于WT/PCA的自适应神经网络人脸识别方法  

Self-adaptive Neural Network to Face Recognition Based on Wavelet Transformation and Principal Component Analysis

作　　者：黄贤武[1] 仲兴荣[1] 王加俊[1] 

机构地区：[1]苏州大学电子信息学院,江苏苏州215021

出　　处：《计算机应用》2003年第6期1-3,6,共4页journal of Computer Applications

摘　　要：提出一种基于小波变换和自适应人工神经网络的PCA人脸识别方法。该方法首先在预处理中对图像进行光照强度的补偿,然后用小波变换的方法提取人脸图像的低频子带,再用PCA方法提取特征分量,并用BP人工神经网络进行训练和识别。此算法将PCA优化的特征抽取与神经网络的自适应性相结合,取得了较高的识别率和优良的抗噪声性能。A face recognition method of principal component analysis(PCA) based on wavelet transformation(WT) and selfadaptive artificial neural network is proposed. After compensating illumination intensity of the image in preprocessing, the low frequency subband of face image was extracted through wavelet transformation. Then the feature components were extracted with PCA. Using the features extracted, faces were recognized by selfadaptive artificial neural network. Owing to the combination of the PCA optimized feature extraction and the selfadaption of neural network, the recognition ratio and robustness were greatly improved.

关 键 词：人胜识别 光照补偿 小波变换 BP人工神经网络 PCA 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]