基于PCA的深度信念网的手势识别研究
|
摘要
针对现有的手势识别均采用有监督模型进行特征提取和识别的现状,提出一种基于PCA的深度信念网(DBN)的半监督的手势特征提取与识别方法。运用所提方法进行了大量的实验,证明该方法与直接将图片输入到DBN网络相比,可以有效降低DBN的训练时间,并且识别率也有所提高;并且该方法与传统的有监督的SVM的手势识别方法相比,训练时间大幅度减少而识别率也有很大的提升。最后,对该方法进行了鲁棒性验证,经过大量实验,证明了其具有很强的鲁棒性。
In contrast to the supervised feature extraction method adopted in gesture recognition,this paper proposed a semi-supervised gesture feature extraction and recognition method based on PCA and deep belief network( DBN). The experimental results show that the proposed method can reduce the training time of DBN and improve the recognition rate compared with the direct input the pictures to the DBN network.Furthermore,comparing with the traditional supervised SVM gesture recognition,a significant reduction in training time and the recognition rate is also greatly improved in this paper. Finally,the robustness of the proposed method is verified by a lot of experiments,which prove that the proposed method has strong robustness.
In contrast to the supervised feature extraction method adopted in gesture recognition,this paper proposed a semi-supervised gesture feature extraction and recognition method based on PCA and deep belief network( DBN). The experimental results show that the proposed method can reduce the training time of DBN and improve the recognition rate compared with the direct input the pictures to the DBN network.Furthermore,comparing with the traditional supervised SVM gesture recognition,a significant reduction in training time and the recognition rate is also greatly improved in this paper. Finally,the robustness of the proposed method is verified by a lot of experiments,which prove that the proposed method has strong robustness.
引文
[1]刘江华,程君实,陈佳品.基于视觉的动态手势识别及其在仿人机器人交互中的应用[J].机器,2002,24(3):46-51.
[2]常亚南.基于HMM的动态手势识别[D].广州:华南理工大学,2012.
[3]蔡芝蔚,吴淑燕,宋俊锋.基于SVM和组合优化模型的手势识别[J].系统仿真学报,2016,8(6):12-17.
[4]HINTON G E.Learning multiple layers of representation[C].Trends in Cognitive Sciences,2007:428-434.
[5]操小文,薄华.基于卷积神经网络的手势识别[J].微型机与应用,2016,35(9):55-57,61.
[6]HINTON G E.Training products of experts by minimizing contrastive divergence[J].Neural Computation,2002,14(8):1771-1800.
[7]陶美平,马力,黄文静,等.基于无监督特征学习的手势识别方法[J].微电子学与计算机,2016,1(4):100-103.
[8]CHANG C C,LIN C J.LIBSVM:a library for support vector machines[EB/OL].[2013-11-12](2016-7-30).http://www.csie.ntu.edu.tw/-cj-lin/libsvm/.
[2]常亚南.基于HMM的动态手势识别[D].广州:华南理工大学,2012.
[3]蔡芝蔚,吴淑燕,宋俊锋.基于SVM和组合优化模型的手势识别[J].系统仿真学报,2016,8(6):12-17.
[4]HINTON G E.Learning multiple layers of representation[C].Trends in Cognitive Sciences,2007:428-434.
[5]操小文,薄华.基于卷积神经网络的手势识别[J].微型机与应用,2016,35(9):55-57,61.
[6]HINTON G E.Training products of experts by minimizing contrastive divergence[J].Neural Computation,2002,14(8):1771-1800.
[7]陶美平,马力,黄文静,等.基于无监督特征学习的手势识别方法[J].微电子学与计算机,2016,1(4):100-103.
[8]CHANG C C,LIN C J.LIBSVM:a library for support vector machines[EB/OL].[2013-11-12](2016-7-30).http://www.csie.ntu.edu.tw/-cj-lin/libsvm/.