基于PCA和ICA的人脸识别

详细信息

Face recognition based on PCA and ICA

摘要: 提出利用主成分分析(PCA)和独立成分分析(ICA)相结合的方法对人脸进行识别。首先对预处理后的图像进行降维,即利用PCA算法对图像进行去二阶相关和降维处理,然后再利用ICA算法获得人脸影像独立基成分,利用人脸影像独立基来构造一子空间,最后利用待识别图像在这个空间上的投影系数进行人脸识别。从两个不同的数据集,将传统的PCA人脸识别算法和提出的人脸识别算法进行比较。从实验数据结果看,提出的PCA和ICA结合人脸识别算法优于传统的PCA人脸识别算法。
- 主成分分析 /
- 独立成分分析 /
- 人脸识别 /
- 特征脸 /
- 独立影像基
Abstract: This paper proposes the face recognition method based on principle component analysis (PCA) and independent component analysis (ICA). PCA and ICA are both multivariable data statistics.In order to reduce dimension and sec order correlation,firstly,the image data can be processed by PCA,and then the independent image basis can be got using ICA,finally,face recognition is processed in the subspace. In our experiments,the proposed methods have been successfully evaluated using two different datasets,and PCA and ICA method are compared.The experimental results show that ICA face recognition method is superior to PCA method.
- principle component analysis(PCA) /
- independent component analysis(ICA) /
- face recognition /
- eigenface /
- independent basis image

[1]	BARTLETT M S.Face image analysis by unsupervised learning and redundancy reduction. Ph.D Thesis of University of California,1998.27～37.
[2]	SIROVICH L,KIRBY M.J O S A,1987,4(3):519～524.
[3]	TURK M A,PENTLAND A P.J Cognitive Neurosci,1994,3(1):71～86.
[4]	COMON P.Signal Processing,1994,36(3):287～314.
[5]	HYVARINEN A.IEEE Trans on Neural Networks,1999,10(3):626～634