Advertisement

Fast 6D object pose refinement in depth images

  • Haoruo ZhangEmail author
  • Qixin Cao
Article
  • 42 Downloads

Abstract

Recovering 6D object pose has gained much focus, because of its application in robotic intelligent manipulation to name but a few. This paper presents an approach for 6D object pose refinement from noisy depth images obtained from a consumer depth sensor. Compared to the state of the art aimed at the same goal, the proposed method has high precision, high robustness to partial occlusions and noise, low computation cost and fast convergence. This is achieved by using an iterative scheme that only employs Random Forest to minimize a cost function of object pose which can quantify the misalignment between the ground truth and the estimated one. The random forest in our algorithm is learnt only using synthetic depth images rendered from 3D model of the object. Several experimental results show the superior performance of the proposed approach compared to ICP-based algorithm and optimization-based algorithm, which are generally used for 6D pose refinement in depth images. Moreover, the iterative process of our algorithm can be much faster than the state of the art by only using one CPU core.

Keywords

6D pose estimation Object pose refinement Depth images Random forest Fast convergence 

Notes

Acknowledgments

This work has been supported by National Natural Science Foundation of China (Grant No. 61673261).

References

  1. 1.
    Collet A, Martinez M, Srinivasa SS (2011) The MOPED framework: object recognition and pose estimation for manipulation. Int J Robot Res 30(10):1284–1306CrossRefGoogle Scholar
  2. 2.
    Savarese S, and Fei-Fei L (2007) 3D generic object categorization, localization and pose estimation. In: International Conference on Computer Vision, pp. 1–8Google Scholar
  3. 3.
    Besl PJ, Mckay ND (1992) A method for registration of 3-D shapes. IEEE Trans Pattern Anal Mach Intell 14(2):239–256CrossRefGoogle Scholar
  4. 4.
    Rusinkiewicz S, Levoy M (2001) Efficient variants of the ICP algorithm. In: IEEE international conference on 3D digital imaging and modeling, pp 145–152CrossRefGoogle Scholar
  5. 5.
    Fisher R. (2001) Projective ICP and stabilizing architectural augmented reality overlays. In: International symposium on virtual and augmented architecture, pp. 69–80Google Scholar
  6. 6.
    Zabulis X, Lourakis M, and Koutlemanis P. (2015) 3D object pose refinement in range images. In: International conference on computer vision systems, pp. 263–274Google Scholar
  7. 7.
    Zabulis X, Lourakis M, Koutlemanis P (2016) Correspondence-free pose estimation for 3D objects from noisy depth data. Vis Comput:1–19Google Scholar
  8. 8.
    Tan D J, Ilic S (2014) Multi-forest tracker: a chameleon in tracking. In: IEEE Conference on Computer Vision and Pattern Recognition, pp. 1202–1209Google Scholar
  9. 9.
    Tan D J, Tombari F, Ilic S, et al. (2015) A versatile learning-based 3D temporal tracker: scalable, robust, online. In: IEEE International Conference on Computer Vision, pp. 693–701Google Scholar
  10. 10.
    Frome A, Huber D, Kolluri R, et al. (2004) Recognizing objects in range data using regional point descriptors. In: European conference on computer vision, pp. 224–237Google Scholar
  11. 11.
    Zaharescu A, Boyer E, Varanasi K, et al. (2009) Surface feature detection and description with applications to mesh matching. In: IEEE Conference on Computer Vision and Pattern Recognition, pp. 373–380Google Scholar
  12. 12.
    Rusu R B, Blodow N, Beetz M (2009) Fast point feature histograms (FPFH) for 3D registration. In: IEEE International Conference on Robotics and Automation, pp. 3212–3217Google Scholar
  13. 13.
    Rusu R B, Bradski G, Thibaux R, et al. (2010) Fast 3d recognition and pose using the viewpoint feature histogram. In: IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 2155–2162Google Scholar
  14. 14.
    Bronstein M M, Kokkinos I (2010) Scale-invariant heat kernel signatures for non-rigid shape recognition. In: IEEE Conference on Computer Vision and Pattern Recognition pp. 1704–1711Google Scholar
  15. 15.
    Drost B, Ulrich M, Navab N, et al. (2010) Model globally, match locally: efficient and robust 3D object recognition. In: IEEE Conference on Computer Vision and Pattern Recognition, pp. 998–1005Google Scholar
  16. 16.
    Hinterstoisser S, Lepetit V, Rajkumar N, et al. (2016) Going further with point pair features. In: European conference on computer vision, pp. 834–848Google Scholar
  17. 17.
    Darom T, Keller Y (2012) Scale-invariant features for 3-D mesh models. IEEE Trans Image Process 21(5):2758–2769MathSciNetCrossRefGoogle Scholar
  18. 18.
    Choi C, Christensen H I (2012) 3D pose estimation of daily objects using an RGB-D camera. In: IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 3342–3349Google Scholar
  19. 19.
    Hinterstoisser S, Lepetit V, Ilic S, et al. (2012) Model based training, detection and pose estimation of texture-less 3d objects in heavily cluttered scenes. In: Asian conference on computer vision, pp. 548–562Google Scholar
  20. 20.
    Hinterstoisser S, Cagniart C, Ilic S et al (2012) Gradient response maps for real-time detection of textureless objects. IEEE Trans Pattern Anal Mach Intell 34(5):876–888CrossRefGoogle Scholar
  21. 21.
    Rios-Cabrera R, Tuytelaars T. (2014) Discriminatively trained templates for 3D object detection: a real time scalable approach. In: IEEE international conference on computer vision, pp. 2048–2055Google Scholar
  22. 22.
    Kehl W, Tombari F, Navab N, et al. (2016) Hashmod: a hashing method for scalable 3D object detection. In: British machine vision conference, pp. 1–12Google Scholar
  23. 23.
    Cai H, Werner T, Matas J (2013) Fast detection of multiple Textureless 3-D objects. In: International Conference on Computer Vision Systems, pp. 103–112Google Scholar
  24. 24.
    Hodan T, Zabulis X, Lourakis M, et al. (2015) Detection and fine 3D pose estimation of texture-less objects in RGB-D images. In: IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 4421–4428Google Scholar
  25. 25.
    Zhang H, Cao Q (2017) Texture-less object detection and 6D pose estimation in RGB-D images. Robot Auton Syst 95:64–79CrossRefGoogle Scholar
  26. 26.
    Bradski G, Bradski G, Xu B X, et al. (2010) Depth-encoded hough voting for joint object detection and shape recovery. In: European Conference on Computer Vision, pp. 658–671Google Scholar
  27. 27.
    Brachmann E, Krull A, Michel F, et al. (2014) Learning 6D object pose estimation using 3D object coordinates. In: European conference on computer vision, pp. 536–551Google Scholar
  28. 28.
    Krull A, Brachmann E, Michel F, et al. (2015) Learning analysis-by-synthesis for 6D pose estimation in RGB-D images. In: International Conference on Computer Vision, pp. 954–962Google Scholar
  29. 29.
    Wohlhart P, Lepetit V (2015) Learning descriptors for object recognition and 3D pose estimation. In: IEEE Conference on Computer Vision and Pattern Recognition, pp. 3109–3118Google Scholar
  30. 30.
    Doumanoglou A, Kouskouridas R, Malassiotis S, et al. (2016) Recovering 6D object pose and predicting next-best-view in the crowd. In: IEEE Conference on Computer Vision and Pattern Recognition, pp. 3583–3592Google Scholar
  31. 31.
    Kehl W, Milletari F, Tombari F, et al. (2016) Deep learning of local RGB-D patches for 3D object detection and 6D pose estimation. In: European conference on computer vision, pp. 205–220Google Scholar
  32. 32.
    Zhang H, Cao Q (2017) Combined holistic and local patches for recovering 6D object pose. In: International Conference on Computer Vision Workshops, pp. 2219–2227Google Scholar
  33. 33.
    Kehl W, Manhardt F, Tombari F, et al. (2017) SSD-6D: making RGB-based 3D detection and 6D pose estimation great again. In: International Conference on Computer Vision, pp. 1530–1538Google Scholar
  34. 34.
    Michel F, Kirillov A, Brachmann E, et al. (2016) Global hypothesis generation for 6D object pose estimation. In: IEEE conference on computer vision and pattern recognition, pp. 115–124Google Scholar
  35. 35.
    Porzi L, Penate-Sanchez A, Ricci E, et al. (2017) Depth-aware convolutional neural networks for accurate 3D pose estimation in RGB-D images. In: IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 5777–5783Google Scholar
  36. 36.
    Jafari OH, Mustikovela SK, Pertsch K, et al. (2017) The Best of BothWorlds: Learning Geometry-based 6D Object Pose Estimation. arXiv preprint arXiv:1712.01924Google Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2019

Authors and Affiliations

  1. 1.State Key Lab of Mechanical Systems and VibrationShanghai Jiao Tong UniversityShanghaiChina

Personalised recommendations