Skip to main content
SpringerLink
Log in

Class specific nullspace marginal discriminant analysis with overfitting-prevention kernel estimation for hand trajectory recognitions

  • Published:
Multimedia Tools and Applications Aims and scope Submit manuscript

Abstract

Hand trajectories are widely used for gesture recognition, action analysis, and sign language translation. Effective hand trajectory feature extraction facilitates accurate and fast identification. In this paper, we propose class specific nullspace marginal discriminant analysis (CSNMDA) with overfitting-prevention kernel estimation for hand trajectory feature extraction, which enhances the similarity of interested class (positive samples) and differentiating arbitrary samples not belonging to this class (negative samples). To address the intrinsic overfitting problems of discriminant analysis, we formulate a kernel space estimation method, which improves the model generalizability and accelerate the training speed. Besides, maximizing the distances between all negative samples and positive samples without distinction leads to limited discriminant power, especially for samples at the margin. Therefore, marginal discriminant analysis (MDA) is conducted to expand the margin of positive and negative class, achieving superior differentiation. According to the experiment on two public available hand trajectory databases, our method obtains higher accuracy compared with Class Specific Kernel Discriminant Analysis (CSKDA) and the Heterogeneous Orthogonal Class Specific Discriminant Analysis (HNCSDA).

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7

Similar content being viewed by others

Data availability

No new data were created during the study, and two existing datasets were used for the experiments. The Three-cent dataset is publicly available at: https://github.com/davidespano/3cent-dataset. The Gesture database is available from the author, Dr. Ho-Sub Yoon (yoonhs@etri.re.kr), upon reasonable request.

References

  1. Alexandros I (2018) Probabilistic class-specific discriminant analysis, arXiv: 1812.05980

  2. Azar SG, Hadi S (2020) Trajectory-based recognition of dynamic Persian sign language using hidden Markov model. Comput Speech Lang 61:101053

    Article  Google Scholar 

  3. Beh J, David H, Hanseok K (2014) Rule-based trajectory segmentation for modeling hand motion trajectory. Pattern Recogn 47(4):1586–1601

    Article  Google Scholar 

  4. Belhumeur PN, Hespanha JP, Kriegman DJ (1997) Eigenfaces vs. Fisherfaces: recognition using class specific linear projection. IEEE Trans Pattern Anal Mach Intell 19(7):711–720

    Article  Google Scholar 

  5. Cao G, Iosifidis A, Gabbouj M (2017) Neural class-specific regression for face verification. IET Biom 7(1):63–70

    Article  Google Scholar 

  6. Caputo FM, Pietro P, Alessandro C et al (2018) Comparing 3D trajectories for simple mid-air gesture recognition. Comput Graph 73:17–25

    Article  Google Scholar 

  7. Chen L, Hong-Yuan ML, Ming-Tat K et al (2000) A new LDA-based face recognition system which can solve the small sample size problem. Pattern Recogn 33(10):1713–1726

    Article  Google Scholar 

  8. Cheng H, Zhongjun D, Zicheng L et al (2016) An image-to-class dynamic time warping approach for both 3D static and trajectory hand gesture recognition. Pattern Recogn 55:137–147

    Article  Google Scholar 

  9. Dong H, Ping L, Chunping L et al (2018) Person re-identification by kernel null space marginal fisher analysis. Pattern Recogn Lett 107:66–74

    Article  Google Scholar 

  10. Everingham M, Luc VG, Christopher KIW et al (2010) The Pascal visual object classes (VOC) challenge. Int J Comput Vis 88(2):303–338

    Article  Google Scholar 

  11. Fenglin L, Wei Z, Chengzhi Y et al (2019) Kinect-based hand gesture recognition using trajectory information, hand motion dynamics and neural networks. Artif Intel Rev 52(1):563–583

    Article  Google Scholar 

  12. Goudelis G, Zafeiriou S, Tefas A et al (2007) Class-specific kernel-discriminant analysis for face verification. IEEE Trans Inf Forensic Secur 2(3):570–587

    Article  Google Scholar 

  13. He Z, Wu M, Zhao X, Zhang S, Tan J (2021) Representative null space LDA for discriminative dimensionality reduction. Pattern Recogn 111:107664

    Article  Google Scholar 

  14. Iosifidis A, Gabbouj M (2016) Scaling up class-specific kernel discriminant analysis for large-scale face verification. IEEE Trans Inf Forensic Secur 11(11):2453–2465

    Article  Google Scholar 

  15. Iosifidis A, Moncef G (2017) Class-specific kernel discriminant analysis revisited: further analysis and extensions. IEEE Trans Cybern 47(12):4485–4496

    Article  Google Scholar 

  16. Iosifidis A, Moncef G, Petri P (2015) Class-specific nonlinear projections using class-specific kernel Spaces. 2015 Ieee trustcom/bigdatase/ispa, Helsinki, Finland, pp 17–24

  17. Iosifidis A, Anastasios T, Ioannis P (2015) Class-specific reference discriminant analysis with application in human behavior analysis. IEEE Tran Human-Mach Syst 45(3):315–326

    Article  Google Scholar 

  18. Jia Z, Zeng Y, Zhang Y, Liang W (2020) Local class-specific discriminant analysis with variable weighting and its application in fault diagnosis. Comput Chem Eng 141:107023

    Article  Google Scholar 

  19. Kwak N (2013) Nonlinear projection trick in kernel methods: an alternative to the kernel trick. IEEE Trans Neural Netw Learn Syst 24(12):2113–2119

    Article  Google Scholar 

  20. Li K, Wu G (2022) Randomized approximate class-specific kernel spectral regression analysis for large-scale face verification. Mach Learn 111(6):2037–2091

    Article  MathSciNet  MATH  Google Scholar 

  21. Marin G, Fabio D, Pietro Z (2014) Hand gesture recognition with leap motion and Kinect Devices, IEEE international conference on image processing (ICIP), Paris, France, pp 1565–1569

  22. Mika S, Ratsch G, Weston J et al (1999) ‘Fisher discriminant analysis with kernels,’ Neural Networks for Signal Processing IX: Proceedings of the 1999 IEEE Signal Processing Society Workshop, Madison, USA, pp 41–48

  23. Park CH, Haesun P (2005) Nonlinear discriminant analysis using kernel functions and the generalized singular value decomposition. SIAM J Matrix Anal Appl 27(1):87–102

    Article  MathSciNet  MATH  Google Scholar 

  24. Raitoharju J, Alexandros I (2019) Null space analysis for class-specific discriminant learning, arXiv:1908.04562

  25. Shotton J, Fitzgibbon A, Cook M et al (2011) Real-time human pose recognition in parts from single depth Images, IEEE conference on computer vision and pattern recognition(CVPR), Colorado, USA, pp 1297–1304

  26. Singh S, Chetan A, CVJ (2017) Trajectory aligned features for first person action recognition. Pattern Recogn 62:45–55

    Article  Google Scholar 

  27. Starner TJW, Pentland A (1998) Real-time American sign language recognition using desk and wearable computer based video. IEEE Trans Pattern Anal Mach Intell 20(12):1371–1375

    Article  Google Scholar 

  28. Tang J, Hong C, Yang Z et al (2018) Structured dynamic time warping for continuous hand trajectory gesture recognition. Pattern Recogn 80:21–31

    Article  Google Scholar 

  29. Wang C, Peng G, De Baets B (2022) Class-specific discriminative metric learning for scene recognition. Pattern Recogn 126:108589

    Article  Google Scholar 

  30. Xiao, R, Pengfei, S (2008) Semi-supervised marginal discriminant analysis based on qr decomposition, 2008 19th international conference on pattern recognition, Tampa, USA, pp 1–4

  31. Yan S, Dong X, Benyu Z et al (2007) Graph embedding and extensions: a general framework for dimensionality reduction. IEEE Trans Pattern Anal Mach Intell 29(1):40–51

    Article  Google Scholar 

  32. Yang C, David KH, Hanseok K (2017) Continuous hand gesture recognition based on trajectory shape information. Pattern Recogn Lett 99:39–47

    Article  Google Scholar 

  33. Yoon H, Jung S, Younglae JB et al (2001) Hand gesture recognition using combined features of location, angle and velocity. Pattern Recogn 34(7):1491–1501

    Article  MATH  Google Scholar 

Download references

Acknowledgments

This research was supported by Zhejiang Provincial Natural Science Foundation of China(LY21E050021), the National Natural Science Foundation of China (51775498, 51775497) and Zhejiang Province Public Welfare Technology Application Research Project (LGG19E050019).

Author information

Author notes

  1. Xinyue Zhao and Gan Gao contributed equally to this work.

Authors and Affiliations

  1. School of Mechanical Engineering, Zhejiang University, Hangzhou, 310027, China

    Xinyue Zhao, Gan Gao & Zaixing He

  2. The State Key Lab of Fluid Power & Mechatronic Systems, Zhejiang University, Hangzhou, 310027, China

    Xinyue Zhao & Zaixing He

  3. Zhejiang Institute of Mechanical and Electrical Engineering, Hangzhou, 310053, China

    Yongfeng Lv

Authors
  1. Xinyue Zhao
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Gan Gao
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Zaixing He
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Yongfeng Lv
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Zaixing He.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhao, X., Gao, G., He, Z. et al. Class specific nullspace marginal discriminant analysis with overfitting-prevention kernel estimation for hand trajectory recognitions. Multimed Tools Appl 82, 46293–46311 (2023). https://doi.org/10.1007/s11042-023-15709-y

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11042-023-15709-y

Keywords

Search

Navigation