摘要
针对利用ICA提取人脸特征时需要将人脸图像转换成向量,导致空间维数很高以及不能准确地估计特征等问题,提出了一种新的独立子空间人脸识别算法——块独立成分分析(B-ICA)。和经典的ICA相比,B-ICA算法把人脸图像划分成一些互不重叠的子块,然后把每个子块转换成向量,看成是低维空间中的训练点(训练向量)。因此在B-ICA算法中,样本的维数比ICA算法中样本的维数低,降低了维数灾难(即样本的训练个数远小于样本的维数)造成的错误识别率。在Yale和AR数据库上进行了大量仿真实验,实验结果表明B-ICA算法的识别率比ICA和其他一些子空间算法的识别率高。
This paper presented a subspace algorithm called Block Independent Component Analysis （B-ICA） for face recognition. Unlike the traditional ICA, in which the whole face image is transformed into a vector before calculating the independent components （ICs）, B-ICA partitions the facial images into blocks and stretches the block to a vector, which is taken as the training vector. Since the dimensionality of the training vector in B-ICA is much smaller than that in traditional ICA, it can reduce the face recognition error caused by the dilemma in ICA, i.e. the number of available training samples is greatly less than that of the dimension of training vector. Extensive experiments are performed on the well-known Yale and AR databases to validate the proposed method and the experimental results show that the B-ICA achieves higher recognition accuracy than ICA and other existing subspace methods.
出处
《计算机应用》
CSCD
北大核心
2007年第9期2091-2094,共4页
journal of Computer Applications
关键词
独立成分分析
特征提取
人脸识别
Independent Component Analysis （ICA）
feature extraction
face recognition
