步态与人脸融合的身份识别方法

Research on Human Identification Based on Gait and Face

为了保持步态识别的优势,克服单一生物特征识别的不足,提高远距离的身份识别率,提出了一种步态与侧面人脸在特征层上融合识别方法.首先利用二向图像矩阵主成分分析,对步态能量图和侧面人脸图分别进行特征提取与降维处理,得到初始特征矩阵,并将得到的初始特征矩阵进行矢量化、特征组合,获得组合特征向量.然后利用多重判别分析法对组合特征向量进行特征融合,获得步态与人脸的融合特征向量,最后采用最近邻法进行身份识别.利用CASIA Dataset B步态数据库对上述方法进行了验证.结果表明,该方法提高了身份识别的正确率,验证了该方法的有效性,为多生物特征识别提供了一种新的方法.

In order to retain the advantage of gait recognition and to overcome the shortcomings of single biometric and increase recognition rate for human recognition at a distance,a novel approach of fusing gait and side face for human recognition at the feature level was presented. Firstly,the feature exaction and dimension reducing was done to Gait Energy Image（ GEI） and Side Face Image,respectively,by Two-Direction Image Matrix based Principal Component Analysis（ 2DIMPCA）,and two original feature vectors were obtained correspondingly,which are vectored and integrated into synthetic feature vectors. Then Multiple Discriminant Analysis（ MDA） was employed on the synthetic feature vectors of gait and side face to obtain fusion features vectors. Finally,the recognition process was implemented on the fusion feature vectors by nearest neighbor（ NN） algorithm. Experiments were implemented on Dataset B of CASIA gait databases and the results showed that the higher correct recognition rate was gained. In the meantime,the validity of the method was demonstrated,and a new approach was supplied for multimodal biometric identification.