摘要
人脸识别是一种重要的生物特征识别技术,它广泛应用于电脑开机、门禁系统、安全监控和重要场所的身份验证。该文提出了一种基于DCT的PCA特征提取方法,该算法先对整个原始人脸图像进行DCT变换得到系数矩阵,再提取包含原始图像大部分信息的少量系数作PCA,提取出特征脸,再进行分类识别。对ORL人脸库的仿真实验表明,该方法优于单独DCT与PCA特征提取的识别方法,并且减少了运算量。
Face recognition is a kind of important biological recognition technology,it is widely used in the computer,the door control system,safety monitoring and important place for authentication.This thesis presents one kind of PCA feature extraction method basing on DCT.The algorithm first to the original face image by DCT transform to get the coefficient matrix.Second,most of the information extracted a small number of coefficients of the original image as containing PCA.Extracting the features of face,then classify. The simulation experiments on ORL face database show that Identification method for extracting DCT alone and PCA features of this method is better than,and reduces the amount of computation.
出处
《电子质量》
2014年第1期34-36,共3页
Electronics Quality
关键词
DCT
PCA
特征提取
图像预处理
Discrete cosine transform
Principal component analysis
Feature extraction
Image prepro- cessing
