摘要: |
针对人脸维度过高和人脸局部特征提取易忽略的问题,提出了一种将多尺度局部二值模式(LBP)算法与深度信念网络(DBN)算法相结合的人脸识别方法。首先采用多尺度LBP算法提取人脸纹理特征,进而将LBP提取的纹理特征作为深度信念网络的输入,最后通过逐层网络训练,得到网络的最优参数,并在ORL人脸库中进行测试,识别率可达95.2%,比使用Gabor小波和主成分分析(PCA)算法的人脸识别高2.6%,说明该算法具有很好的人脸识别能力。 |
关键词: 人脸识别 特征提取 局部二值模式 深度信念网络 受限波尔兹曼机 |
DOI: |
|
基金项目:国家自然科学基金资助项目(61272120);陕西省科技统筹项目(2016KTZDGY02-04-02) |
|
Facial recognition using local binary pattern and deep belief network |
WU Jin,YAN Hui,WANG Jie |
() |
Abstract: |
In view of the large dimensions of facial features and the neglect of the local features extraction,this paper proposes a facial recognition method based on multi-scale local binary pattern(LBP) and deep belief network(DBN) algorithm.Firstly,the texture features are extracted by using multi-scale LBP,then these features are inputed to DBN,and finally the optimal parameters of the network are obtained by layered network training.A test on the ORL faces database is performed and recognition rate reaches 95.2〖WT《Times New Roman》〗%〖WTBZ〗,2.6〖WT《Times New Roman》〗%〖WTBZ〗 higher than that of face recognition using Gabor wavelet and pincipal component analysis(PCA) algorithms,which illustrates that the proposed algorithm has a good face recognition ability. |
Key words: facial recognition feature extraction local binary pattern deep belief network restricted Boltzmann machines |