高级搜索

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于图像协方差无关的增量特征提取方法研究

王肖锋 孙明月 葛为民

王肖锋, 孙明月, 葛为民. 基于图像协方差无关的增量特征提取方法研究[J]. 电子与信息学报, 2019, 41(11): 2768-2776. doi: 10.11999/JEIT181138
引用本文: 王肖锋, 孙明月, 葛为民. 基于图像协方差无关的增量特征提取方法研究[J]. 电子与信息学报, 2019, 41(11): 2768-2776. doi: 10.11999/JEIT181138
Xiaofeng WANG, Mingyue SUN, Weimin GE. An Incremental Feature Extraction Method without Estimating Image Covariance Matrix[J]. Journal of Electronics & Information Technology, 2019, 41(11): 2768-2776. doi: 10.11999/JEIT181138
Citation: Xiaofeng WANG, Mingyue SUN, Weimin GE. An Incremental Feature Extraction Method without Estimating Image Covariance Matrix[J]. Journal of Electronics & Information Technology, 2019, 41(11): 2768-2776. doi: 10.11999/JEIT181138

基于图像协方差无关的增量特征提取方法研究

doi: 10.11999/JEIT181138
基金项目: 国家重点研发计划(2017YFB1303304),天津市科技计划重大专项(17ZXZNGX00110),天津市自然科学重点基金(16JCZDJC30400)
详细信息
    作者简介:

    王肖锋:男,1977年生,博士,讲师,研究方向为发育机器人和机器学习

    孙明月:女,1994年生,硕士生,研究方向为机器人智能学习

    葛为民:男,1968年生,博士,教授,研究方向为机器人智能控制

    通讯作者:

    葛为民 geweimin@tjut.edu.cn

  • 中图分类号: TP391.41

An Incremental Feature Extraction Method without Estimating Image Covariance Matrix

Funds: The National Key R & D Plan of China (2017YFB1303304), The Tianjin Science and Technology Planed Key Project (17ZXZNGX00110), The Tianjin Natural Science Key Foundation (16JCZDJC30400)
  • 摘要: 针对2维主成分分析(2DPCA)算法无法实现在线特征提取及无法体现完整数据结构信息等问题,该文提出一种基于图像协方差无关的增量式2DPCA(I2DPCA)算法。该算法无需对图像协方差矩阵进行特征值分解奇异值分解,复杂度将大为降低,提高了特征提取速度。针对I2DPCA仅提取了横向特征的问题,又提出一种增量式行列顺序2DPCA(IRC2DPCA)算法,该算法对I2DPCA的特征矩阵再次进行纵向特征提取,保留了图像的横向与纵向结构信息,实现了行列两个方向上的特征提取与数据降维。最后,以自建的物块数据集、通用的ORL和Yale人脸数据集分别进行对比实验,结果表明,该文算法在收敛率、分类率及复杂度等性能方面均得到了显著提高,其收敛率达到99%以上,分类率可达97.6%,平均处理速度为29 帧/s,能够满足增量特征提取的实时处理需求。
  • 图  1  部分物块样本图像

    图  2  部分ORL人脸样本图像

    图  3  部分Yale人脸样本图像

    图  4  物块数据集循环多次特征向量的收敛率

    表  1  不同数据集下特征向量收敛率的均值及标准差

    数据集输入次数I2DPCA(均值/标准差)IRC2DPCA(均值/标准差)
    前4个特征向量后4个特征向量前4个特征向量后4个特征向量
    物块$m = 2$0.99442/0.003610.86836/0.178260.98964/0.005730.96564/0.01135
    $m = 5$0.99924/0.000410.95478/0.158180.99824/0.001010.97302/0.01020
    $m = 10$0.99981/0.000100.98453/0.022250.99950/0.000300.99495/0.00322
    ORL$m = 2$0.99991/0.000040.89400/0.067630.99753/0.002030.98175/0.01027
    $m = 5$0.99997/0.000020.91620/0.051540.99951/0.000370.99441/0.00231
    $m = 10$0.99998/0.000010.92833/0.051630.99986/0.000100.99755/0.00074
    Yale$m = 2$0.93732/0.072440.96069/0.030260.99602/0.004410.98308/0.00659
    $m = 5$0.96001/0.046780.98261/0.016090.99924/0.000740.99528/0.00140
    $m = 10$0.97283/0.031800.99041/0.009700.99975/0.000240.99793/0.00088
    下载: 导出CSV

    表  2  物块数据集的最佳分类率

    每类训练
    样本数
    2DPCA
    (120×4)[8]
    RC2DPCA
    (8×8)[11]
    Angle-2DPCA
    (120×4)[12]
    BA2DPCA
    (8×8)[13]
    BDPCA
    (8×8)[10]
    IBDPCA
    (8×8)[17]
    I2DPCA
    (120×4)
    IRC2DPCA
    (8×8)
    L=100.9330.9270.9220.9310.9300.9300.9260.922
    L=250.9570.9600.9580.9610.9590.9590.9620.968
    L=500.9590.9610.9610.9610.9610.9670.9620.972
    L=750.9540.9570.9540.9570.9580.9560.9550.958
    L=1000.9660.9630.9660.9640.9630.9730.9710.976
    下载: 导出CSV

    表  3  ORL数据集的最佳分类率

    每类训练
    样本数
    2DPCA
    (112×4)[8]
    RC2DPCA
    (8×8)[11]
    Angle-2DPCA
    (112×4)[12]
    BA2DPCA
    (8×8)[13]
    BDPCA
    (8×8)[10]
    IBDPCA
    (8×8)[17]
    I2DPCA
    (112×4)
    IRC2DPCA
    (8×8)
    L=10.7440.7280.7420.7250.7280.7080.7360.717
    L=20.8500.8440.8500.8470.8470.8440.8470.841
    L=30.8680.8570.8680.8610.8610.8460.8640.857
    L=40.8880.8960.8880.8920.9040.9040.8830.900
    L=50.9050.9050.9050.9050.9150.9150.9050.925
    下载: 导出CSV

    表  4  Yale数据集的最佳分类率

    每类训练
    样本数
    2DPCA
    (100×4)[8]
    RC2DPCA
    (8×8)[11]
    Angle-2DPCA
    (100×4)[12]
    BA2DPCA
    (8×8)[13]
    BDPCA
    (8×8)[10]
    IBDPCA
    (8×8)[17]
    I2DPCA
    (100×4)
    IRC2DPCA
    (8×8)
    L=10.5600.5600.5600.5600.5600.5530.5730.560
    L=20.7190.7330.7190.7330.7410.7410.7260.726
    L=30.8000.8080.7920.8000.8170.8250.7920.825
    L=40.8570.8760.8570.8760.8760.8760.8570.867
    L=50.8560.8890.8560.8890.8890.8890.8560.889
    下载: 导出CSV

    表  5  物块数据集处理的所需时间对比(s)

    算法L=10L=25L=50L=75L=100
    特征提取分类识别特征提取分类识别特征提取分类识别特征提取分类识别特征提取分类识别
    2DPCA(120×4)[8]2.1525.2799.5686.56132.7457.99069.0659.015123.77910.319
    RC2DPCA(8×8)[11]2.0290.9419.2081.14032.6301.36266.4461.335115.3441.554
    Angle-2DPCA(120×4)[12]4.9911.09715.6041.17493.281.29667.7881.47157.6241.545
    BA2DPCA(8×8)[13]13.1171.17914.2611.19046.1071.20365.1051.25048.0891.233
    BDPCA(8×8)[10]2.3560.9339.4560.96632.9881.24867.5011.414127.391.556
    IBDPCA(8×8)[17]11.8260.86829.0741.05556.7781.22186.1201.408114.8451.604
    I2DPCA(120×4)3.1035.3617.3816.22514.6937.77821.9388.96329.03911.181
    IRC2DPCA(8×8)7.1590.84617.0471.04334.4131.26651.5771.44365.5101.669
    下载: 导出CSV

    表  6  物块数据集处理的所需内存对比(kB)

    算法L=10L=25L=50L=75L=100
    特征提取分类识别特征提取分类识别特征提取分类识别特征提取分类识别特征提取分类识别
    2DPCA(120×4)[8]16.65870.88937.58869.93572.75368.149107.94166.383143.17964.622
    RC2DPCA(8×8)[11]16.73611.35438.19911.42373.97711.227109.86710.805145.76410.612
    Angle-2DPCA(120×4)[12]16.09648.77416.09648.11816.08046.76816.08045.59616.08044.440
    BA2DPCA(8×8)[13]16.08021.64216.08821.14216.10420.5716.1220.50416.09620.036
    BDPCA(8×8)[10]16.37911.34537.04011.42771.66311.243106.43810.768141.16010.620
    IBDPCA(8×8)[17]8.83011.3828.49911.4038.48211.2188.47810.7808.48610.612
    I2DPCA(120×4)8.84770.8978.50369.9358.48668.1418.48666.3018.50364.602
    IRC2DPCA(8×8)8.51111.3508.50711.4198.48611.2438.53610.7978.51110.604
    下载: 导出CSV
  • WANG Hongjuan, HU Jiani, and DENG Weihong. Face feature extraction: a complete review[J]. IEEE Access, 2018, 6: 6001–6039. doi: 10.1109/ACCESS.2017.2784842
    CHAUDHARY G, SRIVASTAVA S, and BHARDWAJ S. Feature extraction methods for speaker recognition: A review[J]. International Journal of Pattern Recognition and Artificial Intelligence, 2017, 31(12): 1–39. doi: 10.1142/S0218001417500410
    SOORA N R and DESHPANDE P S. Review of feature extraction techniques for character recognition[J]. IETE Journal of Research, 2018, 64(2): 280–295. doi: 10.1080/03772063.2017.1351323
    NEIVA D H and ZANCHETTIN C. Gesture recognition: A review focusing on sign language in a mobile context[J]. Expert Systems with Applications, 2018, 103: 159–183. doi: 10.1016/j.eswa.2018.01.051
    陈小龙, 关键, 于晓涵, 等. 基于短时稀疏时频分布的雷达目标微动特征提取及检测方法[J]. 电子与信息学报, 2017, 39(5): 1017–1023. doi: 10.11999/JEIT161040

    CHEN Xiaolong, GUAN Jian, YU Xiaohan, et al. Radar micro-Doppler signature extraction and detection via short-time sparse time-frequency distribution[J]. Journal of Electronics &Information Technology, 2017, 39(5): 1017–1023. doi: 10.11999/JEIT161040
    ISLAM S, ANAND S, HAMID J, et al. Comparing the performance of linear and nonlinear principal components in the context of high-dimensional genomic data integration[J]. Statistical Applications in Genetics and Molecular Biology, 2017, 16(3): 199–216. doi: 10.1515/sagmb-2016-0066
    BRO R and SMILDE A K. Principal component analysis[J]. Analytical Methods, 2014, 6(9): 2812–2831. doi: 10.1039/c3ay41907j
    YANG Jian, ZHANG David, FRANGI A F, et al. Two-dimensional PCA: a new approach to appearance-based face representation and recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2004, 26(1): 131–137. doi: 10.1109/TPAMI.2004.1261097
    ZHANG Daoqiang and ZHOU Zhihua. (2D)2PCA: Two-directional two-dimensional PCA for efficient face representation and recognition[J]. Neurocomputing, 2005, 69(1-3): 224–231. doi: 10.1016/j.neucom.2005.06.004
    ZUO Wangmeng, ZHANG David, and WANG Kuanquan. Bidirectional PCA with assembled matrix distance metric for image recognition[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics) , 2006, 36(4): 863–872. doi: 10.1109/TSMCB.2006.872274
    YANG Wankou, SUN Changyin, and RICANEK K. Sequential Row–Column 2DPCA for face recognition[J]. Neural Computing and Applications, 2012, 21(7): 1729–1735. doi: 10.1007/s00521-011-0676-5
    GAO Quanxue, MA Lan, LIU Yang, et al. Angle 2DPCA: A new formulation for 2DPCA[J]. IEEE Transactions on Cybernetics, 2018, 48(5): 1672–1678. doi: 10.1109/TCYB.2017.2712740.
    ZHOU Shuisheng and ZHANG Danqing. Bilateral angle 2DPCA for face recognition[J]. IEEE Signal Processing Letters, 2019, 26(2): 317–321. doi: 10.1109/LSP.2018.2889925
    DIAZ-CHITO K, FERRI F J, and HERNÁNDEZ-SABATÉ A. An overview of incremental feature extraction methods based on linear subspaces[J]. Knowledge-Based Systems, 2018, 145: 219–235. doi: 10.1016/j.knosys.2018.01.020
    WENG Juyang, ZHANG Yilu, and HWANG Weyshiuan. Candid covariance-free incremental principal component analysis[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2003, 25(8): 1034–1040. doi: 10.1109/TPAMI.2003.1217609
    王肖锋, 张明路, 刘军. 基于增量式双向主成分分析的机器人感知学习方法研究[J]. 电子与信息学报, 2018, 40(3): 618–625. doi: 10.11999/JEIT170561

    WANG Xiaofeng, ZHANG Minglu, and LIU Jun. Robot perceptual learning method based on incremental bidirectional principal component analysis[J]. Journal of Electronics &Information Technology, 2018, 40(3): 618–625. doi: 10.11999/JEIT170561
    谢自强, 葛为民, 王肖锋, 等. 发展型机器人实时特征提取方法研究[J]. 机器人, 2017, 39(2): 189–196. doi: 10.13973/j.cnki.robot.2017.0189

    XIE Ziqiang, GE Weimin, WANG Xiaofeng, et al. Real time feature extraction method of developmental robot[J]. Robot, 2017, 39(2): 189–196. doi: 10.13973/j.cnki.robot.2017.0189
    REN Chuanxian and DAI Daoqing. Incremental learning of bidirectional principal components for face recognition[J]. Pattern Recognition, 2010, 43(1): 318–330. doi: 10.1016/j.patcog.2009.05.020
    曹向海, 刘宏伟, 吴顺君. 一种有效的增量BDPCA算法[J]. 系统仿真学报, 2008, 20(20): 5530–5533. doi: 10.16182/j.cnki.joss.2008.20.041

    CAO Xianghai, LIU Hongwei, and WU Shunjun. A kind of efficient incremental BDPCA algorithm[J]. Journal of System Simulation, 2008, 20(20): 5530–5533. doi: 10.16182/j.cnki.joss.2008.20.041
    文颖, 施鹏飞. 一种基于共同向量结合2DPCA的人脸识别方法[J]. 自动化学报, 2009, 35(2): 202–205. doi: 10.3724/SP.J.1004.2009.00202

    WEN Ying and SHI Pengfei. An approach to face recognition based on common vector and 2DPCA[J]. Acta Automatica Sinica, 2009, 35(2): 202–205. doi: 10.3724/SP.J.1004.2009.00202
  • 加载中
图(4) / 表(6)
计量
  • 文章访问数:  3091
  • HTML全文浏览量:  846
  • PDF下载量:  65
  • 被引次数: 0
出版历程
  • 收稿日期:  2018-12-10
  • 修回日期:  2019-05-06
  • 网络出版日期:  2019-05-22
  • 刊出日期:  2019-11-01

目录

    /

    返回文章
    返回