基于快速二维离散余弦变换算法的人脸特征提取  

Face Feature Extraction Based on Fast 2D-DCT Algorithm

作　　者：林景亮[1] 李祖裕[1] 

机构地区：[1]广西师范大学职业技术师范学院,广西桂林541004

出　　处：《计算机与现代化》2013年第5期70-72,共3页Computer and Modernization

基　　金：广西师范大学青年基金资助项目(20121120)

摘　　要：离散余弦变换(DCT)具有图像信号处理后对图像显示效果影响不明显的优点。本文针对离散余弦变换(DCT)系数在人脸识别中的应用,研究不同的DCT块进行编码后其系数对图像质量和识别结果的影响,提出利用快速2D-DCT(二维离散余弦变换)代替普通DCT的人脸特征提取方法。该方法可以使得在对人脸进行平均分割后的图像块处理时间大大减少,从而能够在同样的时间内提取更多的DCT系数用于人脸识别。实验结果表明,该方法快速而且是有效的。Discrete cosine transform （DCT） has an advantage that is the image signal processing without affecting image quality. The paper proposes a new algorithm for face feature extraction. The face image is the first segmented in average, and the fast 2D-DCT algorithm instead of common DCT algorithm is used in the stage of feature vector extraction. The new algorithm has been tested, and the result proves that 2D-DCT can reduce most of time for the stage of face recognition or can extract more DCT coefficient than common DCT algorithm. Experimental results show that the proposed algorithm is fast and effective.

关 键 词：离散余弦变换 快速算法 人脸特征提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]