基于深度自动编码器的机场安检人脸识别系统设计  

Design of Airport Security Checking Facial Recognition System Based on Deep Auto Encoder

作　　者：张宁[1,2] 朱金福[1] 

机构地区：[1]南京航空航天大学民航学院,南京211106 [2]广州民航职业技术学院,广州510403

出　　处：《计算机测量与控制》2015年第2期644-647,共4页Computer Measurement &Control

基　　金：国家自然科学基金项目(71171111)

摘　　要：由于现有的机场安检系统通常只针对行李和旅客携带的违禁物品进行检测,没有考虑到对于正在通缉的罪犯进行监控和核查,设计了一种基于Gabor小波滤波和深度自动编码器的机场安检人脸识别系统;首先,以ATmega128L为处理器核心,SA7111作为模数转换器,MAX7000作为全局逻辑控制单元设计了机场人脸识别系统硬件,然后采用Gabor小波函数作为卷积核函数,在对原始图像分块的基础上进行卷积,采用多个RBM堆叠组成的自动编码器,通过比较差异算法训练RBM从而自动提取人脸特征,最后构建一个三层的BP神经网络,将自动提取的人脸特征作为输入,将图像标签作为输出层,并通过反向传播算法训练网络进行人脸识别;通过部署仿真实验环境对文中方法进行验证,仿真结果表明:文中系统能较为精确地实现人脸识别,与其它方法相比,具有识别率高和收敛速度快的优点。For the given airport security system only aiming at the checking of forbidden object with passenger or in the packet,not con sidering the monitoring or check for being arrested,a facial recognition system based on gabor wavelet and deep auto encoder is designed.Firstly,ATmega128L is used as the processor core,SA7111 is regarded as the analog digit converter,MAX7000 is used as the global logic control unit so the airport facial recognition system hardware is designed,then the gabor wavelet function is used as the convolutional kernel function to convolution the hulks of the original image,multiple RBM is stacked to get the auto encoder,and the CD algorithm is used to train the auto encoder,the facial feature can be obtained from the trained auto encoder.Finally,the three-layer BP network is built,the auto extraction feature is used as the input and the image label is used as the output,and the BP algorithm is used to train the BP network to realize the facial recognition.Through deploying the simulation experiment environment to verify the method in this paper,the experiment shows the method in this paper can accurately realize the facial recognition,and compared with the other methods,it has the higher recognition rate and quicker convergence rate.

关 键 词：人脸识别 GABOR小波 识别率 深度自动编码器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]