Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 56, Issue 2, 021004(2019)

Improved Image Classification Algorithm Based on Principal Component Analysis Network

Xiaohu Zhao1,2, Liangfei Yin1,3, Yanan Zhu4, Peng Liu1,2、*, Xuekui Wang1,3, and Xueru Shen1,3
Author Affiliations
  • 1 National and Local Joint Engineering Laboratory of Internet Application Technology on Mine, Xuzhou, Jiangsu 221008, China
  • 2 Internet of Things Perception Mine Research Centre, China University of Mining and Technology, Xuzhou, Jiangsu 221008, China
  • 3 School of Information and Control Engineering, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
  • 4 Microsoft (China), Beijing 100080, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (16)
    Equations (0)
    References (43)
    Cited By (3)
    Tools
    Paper Information
    Topics
    Figures & Tables(16)
    Structural diagram of PCANet

    Fig. 1. Structural diagram of PCANet

    Download full size
    Schematic of flat neural network

    Fig. 2. Schematic of flat neural network

    Download full size
    Schematic of flat neural network model update

    Fig. 3. Schematic of flat neural network model update

    Download full size
    Overall flow chart

    Fig. 4. Overall flow chart

    Download full size
    Influence of number of first-layer filters on classification effect

    Fig. 5. Influence of number of first-layer filters on classification effect

    Download full size
    Influence of number of second-layer filters on classification effect

    Fig. 6. Influence of number of second-layer filters on classification effect

    Download full size
    Influence of node-number of feature-mapping layer on classification accuracy

    Fig. 7. Influence of node-number of feature-mapping layer on classification accuracy

    Download full size
    Examples in MNIST dataset

    Fig. 8. Examples in MNIST dataset

    Download full size
    Examples in Cifar-10 dataset

    Fig. 9. Examples in Cifar-10 dataset

    Download full size
    Examples in AR dataset

    Fig. 10. Examples in AR dataset

    Download full size
    Examples in FERET dataset

    Fig. 11. Examples in FERET dataset

    Download full size

    • Table 1. Recognition accuracies of numbers in MNIST dataset

      View table

      Table 1. Recognition accuracies of numbers in MNIST dataset

      MethodHSC[28]K-NN-SCM[29]CDBN[30]ConvNet[31]Stochastic poolingConvNet[32]Conv. Maxout+Dropout[33]ScatNet-2(SVMrbf )[34]
      Accuracy /%77.0063.0082.0053.0047.0045.0043.00
      MethodPCANet[35]DNN(Le-Net5)[35]PCANet+AdaBoostPCANet+RandomForestPCANet+DecisionTreePCANet+SVMProposed
      Accuracy /%94.0095.0588.2472.6870.1898.4899.27

    • Table 2. Predictionaccuracies of targets in Cifar-10 dataset

      View table

      Table 2. Predictionaccuracies of targets in Cifar-10 dataset

      MethodTiledCNN[36]ImprovedLCC [37]KDES-A[38]Stochastic poolingConvNet [32]CNN+Spearmint[39]Conv. Maxout+Dropout[33]NIN[40]
      Accuracy /%73.1074.5076.0084.8785.0288.3289.59
      MethodDNN(Le-Net5)PCANet+AdaBoostPCANet+RandomForestPCANet+DecisionTreePCANet+SVMProposed
      Accuracy /%83.3979.6775.2074.0187.5099.01

    • Table 3. Recognition accuracies of faces in AR dataset

      View table

      Table 3. Recognition accuracies of faces in AR dataset

      MethodLBP[41]P-LBP[42]PCANet[35]DNN(Le-Net5)PCANet+AdaBoostPCANet+RandomForestPCANet+DecisionTreePCANet+SVMOurs
      Accuracy /%81.3380.3385.0080.0673.5962.5957.2075.5985.21

    • Table 4. Recognition accuracies of faces in FERET dataset%

      View table

      Table 4. Recognition accuracies of faces in FERET dataset%

      DatasetFbFcDup1Dup2
      LBP[41]93.0051.0061.0050.00
      P-LBP[42]98.0098.0090.0085.00
      DMMA[43]98.1098.5081.6083.20
      POEM[44]99.6099.5088.8085.00
      PCANet+SVM99.0098.9390.7690.27
      DNN (Le-Net5)99.0598.7991.2790.52
      Proposed99.4099.9093.2989.90

    • Table 5. Training efficiency comparison in MNIST dataset

      View table

      Table 5. Training efficiency comparison in MNIST dataset

      MethodAccuracy /%Trainingtime /sTrainingtime ofPCANet /sTrainingtime ofFNN /s
      PCANet+DecisionTree70.1822191804415
      PCANet+RandomForest72.6820541726328
      PCANet+AdaBoost88.2429402143797
      PCANet+SVM98.4836052976729
      DNN(Le-Net5)99.053195
      Proposed99.781739168257
    Tools

    Get Citation

    Copy Citation Text

    Xiaohu Zhao, Liangfei Yin, Yanan Zhu, Peng Liu, Xuekui Wang, Xueru Shen. Improved Image Classification Algorithm Based on Principal Component Analysis Network[J]. Laser & Optoelectronics Progress, 2019, 56(2): 021004

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Set citation alerts for article
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: Jun. 27, 2018

    Accepted: Jul. 30, 2018

    Published Online: Aug. 1, 2019

    The Author Email: Liu Peng (13814538110@163.com)

    DOI:10.3788/LOP56.021004

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology