摘要
针对中低分辨率车型识别问题,建立一种改进的卷积神经网络(CNN)特征融合模型。采取特征融合策略对CNN中的不同低层特征进行融合重复利用。为防止出现过拟合现象,结合网络模型稀疏化的结构,使用数据增强方法优化训练数据。分析和实验结果表明,该模型不仅能产生更具区分性的特征,而且能避免由环境等因素引起的干扰,与传统CNN模型相比,具有更高的识别准确率。
An improved Convolution Neural Network( CNN) feature fusion model is proposed to solve the problem of mid-low resolution vehicle type recognition. The feature fusion strategy is adopted to fuse and reuse different low-level features in CNN. Among them,in order to prevent over-fitting,combined with the sparse structure of the network model,the data enhancement method is used to optimize the training data. Analysis and experimental results show that the model can not only produce more distinguishing features,but also avoid the interference caused by environmental factors to a certain extent. Compared with the traditional CNN model,it has higher recognition accuracy.
引文
[1]苏衡,周杰,张志浩.超分辨率图像重建方法综述[J].自动化学报,2013,39(8):1202-1213.
[2]厉智,孙玉宝,王枫,等.基于深度卷积神经网络的服装图像分类检索算法[J].计算机工程,2016,42(11):309-315.
[3]胡丹,周兴社,许婉君,等.基于深度特征与LBP纹理融合的视觉跟踪[J].计算机工程,2016,42(9):220-225.
[4]ZEILER M D,FERGUS R.Visualizing and understanding convolutional networks[C]//Proceedings of European Conference on Computer Vision.Berlin,Germany:Springer,2013:818-833.
[5]SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[EB/OL].[2017-11-05].https://arxiv.org/pdf/1409.1556.pdf.
[6]SZEGEDY C,LIU W,JIA Y,et al.Going deeper with convolutions[C]//Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2015:1-9.
[7]HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C]//Proceedings of 2016 IEEEConference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2016:770-778.
[8]SERMANET P,EIGEN D,ZHANG X,et al.OverFeat:integrated recognition,localization and detection using convolutional networks[EB/OL].[2017-11-05].https://arxiv.org/pdf/1312.6229.pdf.
[9]ARORA S,BHASKARA A,GE R,et al.Provable bounds for learning some deep representations[EB/OL].[2017-11-05].http://proceedings.mlr.press/v32/arora14.pdf.
[10]卢良锋,谢志军,叶宏武.基于RGB特征与深度特征融合的物体识别算法[J].计算机工程,2016,42(5):186-193.
[11]LCUN Y,BOTTOU L,BENGIO Y,et al.Gradientbased learning applied to document recognition[J].Proceedings of the IEEE,1998,86(11):2278-2324.
[12]LIN M,CHEN Q,YAN S.Network in network[EB/OL].[2017-11-05].https://arxiv.org/pdf/1312.4400.pdf.
[13]JIA Y,SHELHAMER E,DONAHUE J,et al.Caffe:convolutional architecture for fast feature embedding[EB/OL].[2017-11-05].https://arxiv.org/pdf/1408.5093.pdf.
[14]YANG L,LUO P,CHEN C L,et al.A large-scale car dataset for fine-grained categorization and verification[C]//Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2015:3973-3981.
[15]KRAUSE J,STARK M,DENG J,et al.3D object representations for fine-grained categorization[C]//Proceedings of IEEE International Workshop on 3DRepresentation and Recognition.Washington D.C.,USA:IEEE Press,2013:554-561.
[16]DENG J,KRAUSE J,LI F F.Fine-grained crowdsourcing for fine-grained recognition[C]//Proceedings of 2013 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEEPress,2013:580-587.
[17]WANG J,YANG J,YU K,et al.Locality-constrained linear coding for image classification[C]//Proceedings of 2010IEEE Computer Society Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEEComputer Society,2010:3360-3367.
[18]KRAUSE J,GEBRU T,DENG J,et al.Learning features and parts for fine-grained recognition[C]//Proceedings of International Conference on Pattern Recognition.Washington D.C.,USA:IEEE Computer Society,2014:26-33.
[19]PENG X,HOFFMAN J,YU S X,et al.Fine-to-coarse knowledge transfer for low-res image classification[C]//Proceedings of IEEE International Conference on Image Processing.Washington D.C.,USA:IEEE Press,2016:3683-3687.
[20]SALVADOR J,PéREZ-PELLITERO E.Naive Bayes super-resolution forest[C]//Proceedings of IEEEInternational Conference on Computer Vision.Washington D.C.,USA:IEEE Press,2015:325-333.
[21]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of 2014 IEEEConference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2014:580-587.