基于DT-CWT和SVM的人脸图像集成分类方法

Integration Classification Method of Face Images Based on DT-CWT and SVM

利用二元树复小波变换对人脸图像进行5尺度小波分解,并提取每一尺度下6个方向高频子图小波系数模的均值和标准方差组成60维的特征向量表征人脸,然后采用支持向量机的一对一分类算法对ORL人脸图像库进行分类实验,结果表明二元树复小波变换和支持向量机的集成方法能有效提高人脸图像的分类精度.

The dual - tree complex wavelets transform was used to decompose face images with 5 scales, then 60 - dimensional feature vector was generated by computing the mean and standard deviations from wavelet coefficients of six - direction high - frequency subbands of each scale, and the classification experiment was clone by using one vs one algorithm of the support vector machine based on ORL face image database. The tests results show that the method of integrating the dual - tree complex wavelets transform with the support vector machine can effectively improve the classification accuracy of face images.