Depth Estimation of Face Images Based on the Constrained ICA Model

  • Zhan-Li Sun
  • Kin-Man Lam
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6297)


In this paper, we propose a novel and efficient algorithm to reconstruct the 3D structure of a human face from one or a number of its 2D images with different poses. In our proposed algorithm, the rotation and translation process from a frontal-view face image to a non-frontal-view face image is at first formulated as a constrained independent component analysis (cICA) model. Then, the overcomplete ICA problem is converted into a normal ICA problem. The CANDIDE model is also employed to design a reference signal in our algorithm. Moreover, a model-integration method is proposed to improve the depth-estimation accuracy when multiple non-frontal-view face images are available. Experimental results on a real 3D face image database demonstrate the feasibility and efficiency of the proposed method.


Independent Component Analysis Face Image Independent Component Analysis Blind Source Separation Depth Estimation 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Ahlberg, J.: CANDIDE-3 – updated parameterised face. Linkoping University, Lysator LiTH-ISY-R-2325 (2001)Google Scholar
  2. 2.
    Bregler, C., Hertzmann, A., Biermann, H.: Recovering non-rigid 3D shape from image streams. In: Proc. IEEE Comput. Soc. Conf. on Computer Vision and Pattern Recognition, vol. 2, pp. 690–696 (2000)Google Scholar
  3. 3.
    Cootes, T., Taylor, C.: A mixture model for representing shape variation. In: Clark, A.F. (ed.) Proceedings of the 8th British Machine Vision Conference, vol. 1, pp. 110–119. Britich Machine Vision Association Press (1997)Google Scholar
  4. 4.
    Fanany, M.I., Kumazawa, I.: A neural network for recovering 3D shape from erroneous and few depth maps of shaded images. Pattern Recognition Letters 25(4), 377–389 (2004)CrossRefGoogle Scholar
  5. 5.
    Huang, D.S., Min, J.X.: A new constrained independent component analysis method. IEEE Transaction Neural Networks 18(5), 1532–1535 (2007)CrossRefGoogle Scholar
  6. 6.
    Lu, W., Rajapakse, J.C.: Approach and applications of constrained ICA. IEEE Transaction Neural Networks 16(1), 203–212 (2005)CrossRefGoogle Scholar
  7. 7.
    Lu, W., Rajapakse, J.C.: ICA with reference. Neurocomputing 69, 2244–2257 (2006)CrossRefGoogle Scholar
  8. 8.
    Romdhani, S., Vetter, T.: Efficient, robust and accurate fitting of a 3D morphable model. In: IEEE International Conference on Computer Vision, vol. 2(1), pp. 59–66 (2003)Google Scholar
  9. 9.
    Savran, A., Alyüz, N., Dibeklioğlu, H., Çeliktutan, O., Gökberk, B., Sankur, B., Akarun, L.: Bosphorus database for 3D face analysis. In: Schouten, B., Juul, N.C., Drygajlo, A., Tistarelli, M. (eds.) BIOID 2008. LNCS, vol. 5372, pp. 47–56. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  10. 10.
    Sun, Z.L.: An extension of MISEP for post-nonlinear-linear mixture separation. IEEE Transactions on Circuits and Systems Part II: Express Briefs 56(8), 654–658 (2009)CrossRefGoogle Scholar
  11. 11.
    Sun, Z.L., Huang, D.S., Zheng, C.H., Shang, L.: Optimal selection of time lags for TDSEP based on genetic algorithm. Neurocomputing 69(7-9), 884–887 (2006)CrossRefGoogle Scholar
  12. 12.
    Sun, Z.L., Shang, L.: An improved constrained ICA with reference based unmixing matrix initialization. Neurocomputing 73(4-6), 1013–1017 (2010)CrossRefGoogle Scholar
  13. 13.
    Torresani, L., Hertzmann, A., Bregler, C.: Learning non-rigid 3D shape from 2D motion. In: Advances in Neural Information Processing Systems. MIT Press, Cambridge (2003)Google Scholar
  14. 14.
    Zhang, C.Z., Cohen, F.S.: 3-D face structure extraction and recognition from images using 3-D morphing and distance mapping. IEEE Transactions on Image Processing 11(1), 1249–1259 (2002)CrossRefMathSciNetGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2010

Authors and Affiliations

  • Zhan-Li Sun
    • 1
    • 2
  • Kin-Man Lam
    • 1
  1. 1.Centre for Signal Processing, Department of Electronic and Information EngineeringThe Hong Kong Polytechnic University 
  2. 2.Hefei Institute of Intelligent MachinesChinese Academy of Sciences 

Personalised recommendations