摘要
主成分分析(principal component analysis:PCA)已成功用于人脸识别,但基于主成分分析的人脸识别方法需要将图像数据向量化,而向量化后的图像样本维数非常大,计算代价非常高.二维主成分分析(2 di mension principal component analysis:2DPCA)直接处理图像数据,不需要向量化的过程,2DPCA降低了计算复杂度,但是2DPCA与PCA相比,需要存储更多的系数,即要占用更多的存储空间.本文提出了一种基于小波变换和2DPCA的人脸识别方法,可以克服上述缺点,实验结果证明了该方法的有效性.
Principal component analysis(PCA) has been successfully applied to face recognition.However,image data must be converted into vector with high dimension for the PCA based face recognition methods,which requires too much time to extract the principal components.Two dimension principal component analysis(2DPCA) directly process image data without step of vectorization.Compared with PCA based methods,2DPCA based approaches can lower the computational complexity,but much more spaces are need to store the coefficients of 2DPCA.In this paper,based on wavelet transforms(WT) and 2DPCA,an approach of face recognition was proposed,which could overcome the drawback mentioned above.The experimental results confirmed the effectiveness of the proposed method.
出处
《河北大学学报(自然科学版)》
CAS
北大核心
2010年第5期574-579,共6页
Journal of Hebei University(Natural Science Edition)
基金
国家自然科学基金资助项目(60773062)
河北省自然科学基金资助项目(F2010000323
F2008000635)
河北省应用基础研究重点项目(08963522D)
关键词
小波变换
人脸识别
主成分分析
特征脸
特征提取
wavelet transforms
face recognition
principal component analysis
eigenfaces
feature extraction