Journal of the Institute of Electronics Engineers of Korea SP (대한전자공학회논문지SP)

The Institute of Electronics and Information Engineers (대한전자공학회)
권호 표지

3D Reconstruction using the Key-frame Selection from Reprojection Error

카메라 재투영 오차로부터 중요영상 선택을 이용한 3차원 재구성

  • Seo, Yung-Ho (Dept. of Image Engineering, GSAIM, Chung-Ang University) ;
  • Kim, Sang-Hoon (Dept. of Image Engineering, GSAIM, Chung-Ang University) ;
  • Choi, Jong-Soo (Dept. of Image Engineering, GSAIM, Chung-Ang University)
  • 서융호 (중앙대학교 첨단영상대학원 영상공학과) ;
  • 김상훈 (중앙대학교 첨단영상대학원 영상공학과) ;
  • 최종수 (중앙대학교 첨단영상대학원 영상공학과)
  • Published : 2008.01.25
Download PDF
⟨ Previous Next ⟩

Abstract

Key-frame selection algorithm is defined as the process of selecting a necessary images for 3D reconstruction from the uncalibrated images. Also, camera calibration of images is necessary for 3D reconstuction. In this paper, we propose a new method of Key-frame selection with the minimal error for camera calibration. Using the full-auto-calibration, we estimate camera parameters for all selected Key-frames. We remove the false matching using the fundamental matrix computed by algebraic deviation from the estimated camera parameters. Finally we obtain 3D reconstructed data. Our experimental results show that the proposed approach is required rather lower time costs than others, the error of reconstructed data is the smallest. The elapsed time for estimating the fundamental matrix is very fast and the error of estimated fundamental matrix is similar to others.

중요영상 선택 알고리즘은 다수의 비교정 영상으로부터 3차원 재구성을 위해 필수 영상을 선택하는 과정이다. 또한 3차원 재구성을 위해 영상들 사이의 카메라 자동교정(auto-calibration)이 필수적이다. 본 논문은 재구성 오차를 최대한 줄이는 최적의 영상을 선택하는 중요영상 선택 알고리즘을 제안한다. 선택된 중요영상들 사이의 카메라 투영행렬은 카메라 전자동교정(full-auto-calibration)과정을 통하여 추정한다. 정확하게 추정된 카메라 투영행렬로부터 대수학적 유도를 이용하여 기본행렬(fundamental matrix)을 계산하고, 이로부터 잘못된 대응점들을 제거하여 최종적으로 3차원 데이터를 얻는다. 실험 결과는 제안한 중요영상 선택 알고리즘이 다른 알고리즘에 비해 적은 시간이 소요되며, 재구성된 3차원 데이터의 오차가 가장 작았다. 대수학적 유도로부터 얻어낸 기본행렬은 다른 알고리즘에 비해 매우 짧은 시간이 소요 되며 평균 오차는 비슷한 결과를 갖는다.

Keywords

References

  1. A. Watt, F. Policarpo, "The Computer Image", Addison Westley, 1998
  2. K. Ikeuchi, A. Nakazawa, K. Hasegawa, T. Ohishi, "The Great Buddha Project: Modeling Cultural Heritage for VR Systems through Observation", IEEE ACM ISMAR, p. 7, 2003
  3. http://www.cvl.iis.u-tokyo.ac.jp/
  4. http://www.holomatrix.com/
  5. S. Gibson, J. Cook, T. Howard, R. Hubbold, "Accurate Camera Calibration for Off-line, Video-Based Augmented Reality", IEEE ACM ISMAR, pp. 37-46, 2002
  6. J. Repko, M. Pollefeys, "3D Models from Extended Uncalibrated Video Sequences: Addressing Key-frame Selection and Projective Drift", 3DDIM, June 2005
  7. D. Nister, "Frame Decimation for Sturctrue and Motion", In proc. SMILE, Dublin, Ireland, July 2000
  8. Z. Zhang, "Determining the epipolar geometry and its uncertainty: a review", Int. J. of Computer Vision, Vol. 27, No. 2, pp. 161-198, 1998 https://doi.org/10.1023/A:1007941100561
  9. P. H. S. Torr, W. Murray, "The development and comparison of robust methods for estimating the fundamental matrix", Int. J. of Computer Vision, Vol. 34, No. 3, pp. 271-300, 1997
  10. X. Armangue, J. Salvi, "Overall view regarding fundamental matrix estimation", Image and Vision Computing, Vol. 21, No. 2, pp. 205-220, Feb, 2003 https://doi.org/10.1016/S0262-8856(02)00154-3
  11. J. Seo, S. Kim, C. Jho and H. Hong, "3D Estimation and Key Frame selection for Match Move", In Proc. ITC-CSCC, July 2003
  12. T. Thormahlen, H. Broszio, A. Weissenfeld, "Keyframe selection for Camera Motion and Structure Estimation from Multiple Views", LNCS 3021, pp. 523-535, 2004
  13. R. Hartley, A. Zisserman, "Multiple View Geometry", Cambridge Univ. Press, 2000
  14. A. Heyden, K. Astrom, "Euclidean reconstruction from image sequences with varying and unknown focal length and principal point", In Proc. CVPR, pp. 438-443, 1997
  15. B. Triggs, P. McLauchlan, R. Hartley, A. Fitzgibbon, "Bundle Adjustment - a modern synthesis, In Vision algorithms: Theory and practice", LNCS 1883, pp. 298-372, 2000
  16. T. Sato, M. Kanbara, N. ?Yokoya, "Outdoor Scene Reconstruction from Multiple Image Sequences Captured by a Hand-held Video Camera", IEEE Conf. on MFI for Intelligent. Systems, pp. 113-118, July 2003
  17. F. Devernay, O. Faugeras, "From Projective to Euclidean Reconstruction", IEEE ocnf. on CVPR, pp. 264-269, 1996
  18. P. Sturm, B. Triggs, "A Factorization Based Algorithm for Multi-Image Projective Structure and Motion", 4th European Conf. on Computer Vision, pp. 709-720, April 1996
  19. D. Matrinec, T. Pajdla, "Structure from Many perspective Images with Occlusions", In Proc. ECCV, pp. 355-369, May 2002