基于离散余弦变换和支持向量机的人脸识别方法研究

本文实现了一种基于离散余弦变换（DCT）和支持向量机（SVM）的人脸识别方法。该方法利用DCT提取人脸特征向量，输入到SVM分类器中实现人脸的识别．本研究对DCT系数的选取，DCT系数的个数及SVM核参数对识别的影响作了详尽的分析．使用SVM多类分类器的1-1判别策略，实现了一个鲁棒性和实时性均很好的人脸识别方法。实验表明。离散余弦变换技术和支持向量机的结合用于人脸识别。大大提高了识别的速度和识别率。

Research on face recognition system based on DCT and SVM

A novel face recognition method was implemented in this paper, which combined Discrete Cosine Transform (DCT) and Support Vector Machine (SVM). A face image was represented by DCT coefficients, which were then put into SVM classifiers for face recognition. Proper DCT coefficients and support vector kernel function were selected after theoretical and experimental studies. Experimental results show the higher recognizing speed and rate of the proposed method.