基于独立成分分析的人脸图像特征提取与识别

特征提取是人脸识别的关键环节之一。文章首先简述了独立成分分析（ Independent Component Analysis，ICA）的基本模型和原理，介绍了快速独立成分分析FastICA方法特征提取的一般过程。然后给出了FastICA算法中分离矩阵的并行计算算法。最后，利用ORL人脸图像数据库在Matlab环境下进行了仿真实验。实验结果表明，FastICA方法是一种有效的特征提取方法，并讨论了影响分类识别的几个因素。

The Feature Extraction and Face Image Recognition Based on Independent Component Analysis

Feature extraction is one of the key steps of face recognition. It first outlined the basic models and principles of the independent component analysis （ ICA） and described the general process of using fast independ-ent component analysis （ FastICA） for feature extraction. Then gave a parallel computing algorithms of FastICA for separation matrix. At last, it conducted a simulation experiment using ORL face image database in Matlab environ-ment. The experimental results show that, this method of FastICA is an effective method for feature extraction. In addition, it also discussed several factors that affect the classification and recognition in the end.