基于离散小波变换和ICA支持向量机的人脸识别

Facial Recognition Based on Discrete Wavelet Transform and Principal Component Analysis Support Vector Machine

为了实现对具有不同光照、姿势和噪声的人脸进行识别并提高识别精度,设计了一种基于离散小波变换和最小二乘支持向量机的人脸识别方法。首先,采用二维离散小波变换对人脸图像进行压缩和降噪,以提取低频特征信息分量,然后采用快速独立成分分析法ICA对经过离散小波变换后的人脸低频分量进行特征提取,以进一步减少人脸特征向量维数。在获取图像特征向量的基础上,采用径向基函数作为核函数,将训练样本数据输入最小二乘支持向量机进行训练以获得最终的分类模型。在ORL数据库下采用MATLAB仿真工具进行仿真,实验结果表明,该方法能有效地实现对人脸识别,与其他方法相比具有较高的识别精度。

In order to realize facial recognition with different characters such as illumination,posture and noise and improve the recognition precision, a facial recognition method based on discrete wavelet transform and least squares support vector machine is proposed. Firstly ,the discrete wavelet trans- form is used to compress the facial figure and reducing the noise to get the character information component with low frequency,and then the fast inde- pendent component analysis is used to obtain the facial character information with low frequency to reduce the dimension further. Finally, the radius basis function is used as the kernel function,and the training data is input to the least squares support vector machine to get the final recognition model. The simulation experiment is simulated in ORL database with MATLAB tool, and the result shows the method in this paper can realize the facial recognition, and it has big recognition precision compared with the other methods.