摘要
提出一种新的人脸图像特征提取方法,即利用二维经验模态分解方法(BEMD)结合分形维数(Fractal dimension)进行特征量提取,将提取得到的特征量用于人脸识别。该方法将图像通过BEMD算法分解为不同的二维固有模态分量(BIMF),然后将得到的BIMF图像进行分块得到BIMF子区域,对每一个BIMF子区域进行分形盒维数估计,采用BP神经网络作为分类器。实验选用ORL人脸数据库,实验结果表明,用该算法进行特征量提取的人脸识别方法具有理想的识别效果并提高识别系统性能。
A methodology for face recognition based on Bidimensional Empirical Mode Decomposition(BEMD)and fractal box-counting method is proposed.The methodology involves feature extraction of face image using BEMD and fractal box-counting dimension.After the preprocessing procedure,the effective face image is decomposed into2D Intrinsic Mode Function(IMF)components at different spatial frequencies by BEMD.Then the texture features of each intrinsic mode function image are obtained via the box-counting method.To evaluate the efficacy of the proposed method,BPNN used in recognition.The experimental results using the ORL face database show that the proposed method achieves promising results and improve the performance of the recognition system.
作者
陈晓娟
王单卉
CHEN Xiaojuan;WANG Danhui(School of Electronic and Information Engineering, Changchun University of Science and Technology, Changchun 130022, China;School of Information Engineering, Northeast Dianli University, Jilin 132012, China)
出处
《计算机工程与应用》
CSCD
北大核心
2017年第10期177-180,共4页
Computer Engineering and Applications
基金
国家自然科学基金(No.61271115)
关键词
人脸识别
二维经验模态分解
分形维数
特征提取
face recognition
bidimensional empirical mode decomposition
fractal dimension
feature extraction