摘要
研究了使用Gabor滤波和PCA主成分分析法来实现人脸特征提取并使用蚁群优化BP神经网络进行人脸识别的方法.首先,使用Gabor滤波器对预处理后的图像生成不同尺度和方向下的特征向量,然后使用PCA主成分分析法对特征向量进行压缩,为了提高BP神经网络对表情的分类精度和减少训练时间,使用蚁群算法优化BP神经网络的各参数,最后使用优化后的BP神经网络进行训练和人脸识别.仿真实验表明文中的方法能有效地实现对人脸表情进行分类,且较其他方法具有更高的识别率.
The Gabor filter and primary component analysis （PCA） were researched to extract the facial feature and ant colony optimizing （ACO） the BP neural network was used to realize facial recognition. Firstly, facial expression was pre - treated, then the Gabor filter was use to generate the feature vector under different dimension and direction, PCA method was used to reduce the dimension of the final feature. For improving BP neural network classification precision and reducing training time, the ACO algorism was used to optimize the parameters of the BP neural network. The simulation result shows our method realize the classification and has the high recognition rate comparative to other methods.
出处
《湖南科技大学学报（自然科学版）》
CAS
北大核心
2012年第4期69-72,共4页
Journal of Hunan University of Science And Technology：Natural Science Edition
基金
国家自然科学基金(608721042)
江苏省宿迁学院科研基金项目(2012ky17)
关键词
GABOR滤波器
蚁群算法
主成分分析
人脸识别
Gabor filter
ant colony optimizing
primary component analysis
facial recognition
