摘要
本文通过深度卷积网络(Deep Convolutional Neural Networks,DCNN)实现人脸的三维特征和二维特征的提取。对建立的两个DCNN进行训练以及识别测试。将两个DCNN提取的二维人脸图像及人脸深度图的高层抽象特征作为一神经网络(Artificial Neural Network,ANN)的输入,输出结果作为提取的最终特征。实验结果表明,与其他识别方法相比,本文设计的方法在识别正确率上得到了可观的提高。
引文
[1]Krizhevsky A,Sutskever I,Hinton G E.ImageNet classification with deep convolutional neural networks[C].International Conference on Neural Information Processing Systems.Curran Associates Inc.2012:1097-1105.
[2]Erdem C E,Ulukaya S,Karaali A,et al.Combining Haar Feature and skin color based classifiers for face detection[C].IEEEInternational Conference on Acoustics,Speech and Signal Processing.IEEE,2011:1497-1500.
[3]Huang C C,Tsai C Y,Yang H C.An Extended Set of Haarlike Features for Bird Detection Based on AdaBoost[C].Signal Processing,Image Processing and Pattern Recognition-International Conference,Sip 2011,Held As.DBLP,2011:160-169.
[4]Yan L V,Gong Q.Application of weighting 3D-Otsu method in image segmentation[J].Application Research of Computers,2011,28(04):1576-1579.