基于Matlab人脸识别(PCA算法)

摘要

随着科技的发展，人类社会的进步，传统身份识别由于容易遗失，容易被破解已不能起到身份识别作用。人们需要更加安全可靠的身份识别技术。而生物特征的独一无二，不易丢失和被复制的特性很好满足了身份识别的需要。同时随着计算机科学技术和生物医学的发展使得利用生物特征识别成为了可能。在生物特征识别领域，由于人脸识别的操作快速简单，结果直观，准确可靠，不需要人的配合等优点已成为人们关注的焦点。主成分分析（principal component analysis ，PCA）通过提取高维度的人脸图像的主元，使得图像在低维度空间中被处理来降低了图像处理的难度。由于其有效的解决了图像空间维数过高的问题，已经成为人脸识别领域非常重要的理论。本文研究的就是基于PCA的人脸识别算法的实现。

本文按照完整人脸识别流程来分析基于PCA的人脸识别算法实现的性能。首先使用常用的人脸图像的获取方法获取人脸图像。本文为了更好的分析基于PCA人脸识别系统的性能选用了Orl人脸数据库。接下来是人脸图像预处理方法。由于Orl人脸图像质量较好，而且已经做过相应的预处理，所以本文试验中只使用灰度处理。接着使用PCA提取人脸特征，使用奇异值分解定理计算协方差矩阵的特征值和特征向量以及使用最近邻法分类器欧几里得距离来进行人脸判别分类。在实验中我们发现基于PCA的人脸识别系统的识别率很高，而且具有一定鲁棒性，所以基于PCA的人脸识别算法的实现的研究还是有意义。

关键词：人脸识别 PCA算法 奇异值分解定理 欧几里得距离

Abstract

With the development of science and technology, the progress of human society, the traditional identification is easy to lose, easy to be cracked and it has not play an identifiable role. People need a more secure and reliable identification technology. Biometric is unique, easy to lose and replication characteristics of good meet the needs of the identification. With the development of computer science and technology and biomedical makes use of biometric identification has become possible. In the field of biometric identification, face recognition with the advantages of operation is fast and simple, the results are intuitive, accurate and reliable, do not need co-ordination, has become the focus of attention. The principal component analysis (PCA) to extract high dimensional face image of the main element, making the images are processed in low-dimensional space and it reduces the difficulty of image processing. PCA solves effectively the problem of high dimension image space and it has become a very important theory in face recognition field. This paper is in this context of writing from.
In accordance with the full recognition process to analyze the performance of PCA-based face recognition algorithm. The first to use the method of access to commonly used face images for face images. In order to better analysis is based on the performance of the PCA face recognition system selected Essex face database. Next is the face image preprocessing methods. Essex face image quality is better, and have done the appropriate pretreatment, using only gray-scale processing of this trial. Then use the PCA for face feature extraction using singular value decomposition theorem to calculate the covariance matrix of the eigenvalues and eigenvectors, and use the Euclidean distance of the nearest neighbor classifier to the classification of human face discrimination. In the experiment, we found that a high recognition rate of the PCA-based face recognition system, but with a certain robustness, the PCA-based face recognition