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.
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）.
出处
《浙江师范大学学报（自然科学版）》
CAS
2010年第2期128-131,共4页
Journal of Zhejiang Normal University:Natural Sciences
基金
浙江省教育厅重点科研项目(20070440)
关键词
主动表观模型
多角度模型
线性判别分析
主成分分析
active appearance models
multiple-view model
linear discriminant analysis
principal component analysis
