【摘要】 针对传统人脸识别方法的识别率受角度变化影响大的缺点,提出了一种多角度的人脸识别方法.首先用多个不同角度的主动表观模型(AAM)估计人脸旋转角度;然后通过多角度模型合成正面人脸;最后采用线性判别分析法(LDA)获取最佳分类特征,并对其进行识别.实验结果表明:与主成分分析法(PCA)相比,该方法对不同角度的人脸具有更高的识别率.更多还原

【Abstract】 The recognition rate of traditional face recognition method was affected by the view of face changed.To address this problem,a multi-view face recognition method was presented.Firstly,the view of face was evaluated by different angle of active appearance models(AAM).Then the front face was synthesized through the multi-view model.Lastly,the best classification feature was acquired by utilizing the Linear Discriminant Analysis(LDA).The experimental results illustrated that this method had superior recognition rate than Principal Component Analysis(PCA).更多还原