基于改进的独立分量分析的人脸识别方法

将独立分量分析（Independent Component Analysis，ICA）作为人脸特征提取方法。ICA所提取的特征分类能力强、相互独立，对像素间高阶统计特性敏感，并且不易受光照变化的影响。实验结果表明，基于IcA的人脸特征提取方法的识别性能优于特征脸法。针对传统的ICA算法（Informax算法）存在迭代次数多，难收敛，并且需要人工设定步长来调整学习速度的不足，本文采用FastICA作为ICA的快速算法，并将其关键迭代步骤加以改进，减少了耗时的雅可比矩阵求逆的运算次数。所提出的改进的FastICA具有无需人工参与，收敛速度快，迭代次数少的优点。在特征选择方面，本文将遗传算法（Genetie Algorithm，GA）应用到独立分量的选择与优化中，从而在保证较高识别性能的前提下，获得最优的人脸特征子集。

Face Recognition Method Based on Improved Independent Component Analysis

Independent component analysis(ICA) is used as an efficient face feature extraction method.ICA is sensitive to the high-order statistics of the data and finds not-necessarily orthogonal bases,so it can identify and reconstruct high-dimensional face image data better than the principle component analysis(PCA).ICA algorithms are time-consuming and sometimes converge difficultly.A modified fast ICA algorithm is developed,which only needs to compute the Jacobian matrix once in several iterations and achieves the corresponding effect of fast ICA.After obtaining all independent components,a genetic algorithm(GA) is introduced to select optimal independent components(ICs).ICA is compared with the feature extraction method based on PCA.Experimental results show that the modified fast ICA algorithm reduces iteration times and increases the convergence speed.Furthermore,the GA optimizes the recognition performance with least features.The ICA based features extraction method is robust to variations and promising for face recognition.