Skip to main content

Advertisement

SpringerLink
Log in

3D point cloud registration denoising method for human motion image using deep learning algorithm

  • Special Issue Paper
  • Published:
Multimedia Systems Aims and scope Submit manuscript

Abstract

Aiming at the problem of 3D point cloud noise affecting the efficiency and precision of human body 3D reconstruction in complex scenes, a 3D point cloud registration denoising method for human motion image using depth learning algorithm is proposed. First, two Kinect sensors are used to collect the three-dimensional data of the human body in the scene, and the spatial alignment under the Bursa linear model is used to pre-process the background point cloud data. The depth image of the point cloud is calculated, and the depth image pair is extracted by the convolutional neural network. Furthermore, the feature difference of the depth image pair is taken as the input of the fully connected network and the point cloud registration parameter is calculated, and the above operation is performed iteratively until the registration error is less than the acceptable threshold. Then, the improved C-means algorithm is used to remove the outlier, the noise is clustered, and the large-scale outlier noise is removed. Finally, the high-frequency information is processed by the depth data bilateral filtering method. The experimental results show that compared with the traditional bilateral filtering algorithm and fuzzy C-means algorithm, the proposed method can effectively remove noise of different scales and maintain good performance on the basis of maintaining human body features. In the point cloud model of A, B, and C, the average error of the proposed method is lower than that of the traditional bilateral filtering algorithm with 15.7%, 15.9%, and 19.8%, respectively, and it is lower than that of the fuzzy C-means algorithm with 25.8%, 26.9%, and 30.2%, respectively.

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.

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

Similar content being viewed by others

References

  1. Su, T., Wang, W., Lv, Z., et al.: Rapid Delaunay triangulation for randomly distributed point cloud data using adaptive Hilbert curve. Comput. Graph. 54(C), 65–74 (2016)

    Article  Google Scholar 

  2. März, T., Weinmann, A.: Model-based reconstruction for magnetic particle imaging in 2D and 3D. Inverse Prob. Imaging 10(4), 1087–1110 (2017)

    Article  MathSciNet  MATH  Google Scholar 

  3. Clarkson, S., Wheat, J., Heller, B., et al.: Assessment of a Microsoft Kinect-based 3D scanning system for taking body segment girth measurements: a comparison to ISAK and ISO standards. J. Sports Sci. 34(11), 1006–1014 (2016)

    Article  Google Scholar 

  4. Qin, C., Song, A., Wu, C., et al.: Scenario interaction system of rehabilitation training robot based on Unity 3D and Kinect. Chin. J. Sci. Instrum. 38(3), 530–536 (2017)

    Google Scholar 

  5. Pöhlmann, S.T., Harkness, E.F., Taylor, C.J., et al.: Evaluation of Kinect 3D sensor for healthcare imaging. J. Med. Biol. Eng. 36(6), 857–870 (2016)

    Article  Google Scholar 

  6. Yu, Y.L., Zhang, H., Liu, G.H., Shi, J.F.: Kinect depth map pre-processing based on uncertainty evaluation. J. Comput. Appl. 36(3), 541–545 (2016)

    Google Scholar 

  7. Moreno C, Ming L. A progressive transmission technique for the streaming of point cloud data using the Kinect. In: 2018 international conference on computing, networking and communications (ICNC), pp. 593–598. IEEE, Maui, HI, USA (2018)

    Google Scholar 

  8. Zhang, Y., Cong, C., Wu, Q., et al.: A Kinect-based approach for 3D pavement surface reconstruction and cracking recognition. IEEE Trans. Intell. Transport. Syst. 19(99), 1–12 (2018)

    Article  Google Scholar 

  9. Alexiadis, D.S., Chatzitofis, A., Zioulis, N., et al.: An integrated platform for live 3D human reconstruction and motion capturing. IEEE Trans. Circuits Syst. Video Technol. 27(4), 798–813 (2017)

    Article  Google Scholar 

  10. Simo-Serra, E., Torras, C., Moreno-Noguer, F.: 3D human pose tracking priors using geodesic mixture models. Int. J. Comput. Vis. 122(2), 1–21 (2017)

    Article  MathSciNet  Google Scholar 

  11. Chen, H., Ran, L., Lei, W., et al.: Point set surface compression based on shape pattern analysis. Multimed. Tools Appl. 76(20), 20545–20565 (2017)

    Article  Google Scholar 

  12. Rypl, D., Nerad, J.: Volume preserving smoothing of triangular isotropic three-dimensional surface meshes. Adv. Eng. Softw. 101(C), 3–26 (2016)

    Article  Google Scholar 

  13. Lu, X., Liu, X., Deng, Z., et al.: An efficient approach for feature-preserving mesh denoising. Opt. Lasers Eng. 90(9), 186–195 (2017)

    Article  Google Scholar 

  14. Zhang, X., Wan, W., An, X.: Clustering and DCT based color point cloud compression. J. Sig. Process. Syst. 86(1), 41–49 (2017)

    Article  Google Scholar 

  15. Zheng, Y., Li, G., Wu, S., et al.: ‘Guided point cloud denoising via sharp feature skeletons. Vis. Comput. 33(1), 1–11 (2017)

    Article  Google Scholar 

  16. Xie, Z., Liu, J., Pan, C., et al.: A data reduction and ordering algorithm for scattered and layered point cloud. J. Graph. 37(3), 359–366 (2016)

    Google Scholar 

  17. Tao, Y., Li, Y., Wang, Y.Q., et al.: On-line point cloud data extraction algorithm for spatial scanning measurement of irregular surface in copying manufacture. Int. J. Adv. Manuf. Technol. 87(5), 1–15 (2016)

    Article  Google Scholar 

  18. Ide H, Kurita T. Improvement of learning for CNN with ReLU activation by sparse regularization. International Joint Conference on Neural Networks (2017)

  19. Qin, J., Fu, W., Gao, H., et al.: Distributed k-means algorithm and fuzzy c-means algorithm for sensor networks based on multiagent consensus theory. IEEE Trans. Cybern. 47(3), 772–783 (2017)

    Article  Google Scholar 

  20. Galinsky, K., Bhatia, G., Loh, P.R., et al.: Fast principal-component analysis reveals convergent evolution of ADH1B in Europe and East Asia. Am. J. Hum. Genet. 98(3), 456–472 (2016)

    Article  Google Scholar 

  21. Hua, Y., Jiankeng, P., Jianwen, M.O.: Denoising algorithm for bilateral filtered point cloud based on noise classification. J. Comput. Appl. 35(8), 2305–2310 (2015)

    Google Scholar 

  22. Wang, P., Li, W., Ogunbona, P., et al.: RGB-D-based human motion recognition with deep learning: a survey. Comput. Vis. Image Understand. 171(1), 1–22 (2017)

    Article  Google Scholar 

  23. Xiao Q, Chu C. Human motion retrieval based on deep learning and dynamic time warping. International Conference on Robotics & Automation Engineering (2018)

  24. Shao Y, Guo S, Lin S, et al. Human motion classification based on range information with deep convolutional neural network. 2017 4th International Conference on Information Science and Control Engineering (ICISCE) (2017)

  25. Harguess J, Miclat J, Raheema J. Using image quality metrics to identify adversarial imagery for deep learning networks. Geospatial Informatics, Fusion, & Motion Video Analytics VII (2017)

  26. Bag, S., Venkatachalapathy, V.: Ptucha RW (2017) Motion estimation using visual odometry and deep learning localization. Electron. Imaging 19, 62–69 (2017)

    Article  Google Scholar 

  27. Min, W., Cui, H., Rao, H., et al.: Detection of human falls on furniture using scene analysis based on deep learning and activity characteristics. IEEE Access 6(99), 9324–9335 (2018)

    Article  Google Scholar 

  28. Ochi H, Wan W, Yang Y, et al. Deep Learning Scooping Motion using Bilateral Teleoperations. 2018 3rd International Conference on Advanced Robotics and Mechatronics (ICARM) (2018)

  29. Laskey M, Chuck C, Lee J, et al. Comparing human-centric and robot-centric sampling for robot deep learning from demonstrations. IEEE International Conference on Robotics & Automation (2017)

  30. Wu D, Sharma N, Blumenstein M. Recent advances in video-based human action recognition using deep learning: a review. International Joint Conference on Neural Networks (2017)

Download references

Acknowledgements

This work was supported by “Research and application of virtual emergency platform for emergency response of metro project” of Guangdong science and technology project (item number: 2015A030401005); “Human attitude estimation based on visual attention mechanism” of Guangzhou Railway Polytechnic project (item number: GTXYY1614).

Author information

Authors and Affiliations

  1. Educational Technology Center, Guangzhou Railway Polytechnic, Guangzhou, 510430, Guangdong, China

    Qidong Du

Authors
  1. Qidong Du
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Qidong Du.

Additional information

Publisher's Note

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

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Du, Q. 3D point cloud registration denoising method for human motion image using deep learning algorithm. Multimedia Systems 26, 75–82 (2020). https://doi.org/10.1007/s00530-019-00630-y

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00530-019-00630-y

Keywords

Search

Navigation