Belief Propagation Implementation Using CUDA on an NVIDIA GTX 280

  • Yanyan Xu
  • Hui Chen
  • Reinhard Klette
  • Jiaju Liu
  • Tobi Vaudrey
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5866)


Disparity map generation is a significant component of vision-based driver assistance systems. This paper describes an efficient implementation of a belief propagation algorithm on a graphics card (GPU) using CUDA (Compute Uniform Device Architecture) that can be used to speed up stereo image processing by between 30 and 250 times. For evaluation purposes, different kinds of images have been used: reference images from the Middlebury stereo website, and real-world stereo sequences, self-recorded with the research vehicle of the .enpeda.. project at The University of Auckland. This paper provides implementation details, primarily concerned with the inequality constraints, involving the threads and shared memory, required for efficient programming on a GPU.


Shared Memory Belief Propagation Stereo Pair Thread Block Single Instruction Multiple Data 
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.
  2. 2.
    Boykov, Y., Kolmogorov, V.: An experimental comparison of min-cut / max-flow algorithms for energy minimization in vision. IEEE Trans. Pattern Analysis Machine Intelligence 26, 1124–1137 (2004)CrossRefGoogle Scholar
  3. 3.
    Brunton, A., Chang, S., Gerhard, R.: Belief Propagation on the GPU for Stereo Vision. In: Proc. Canadian Conf. Computer Robot Vision, p. 76 (2006)Google Scholar
  4. 4.
    .enpeda.. image sequence analysis test site (EISATS),
  5. 5.
    Felzenszwalb, P.F., Huttenlocher, D.P.: Efficient belief propagation for early vision. Int. J. Computer Vision 70, 41–54 (2006)CrossRefGoogle Scholar
  6. 6.
    Fung, J., Mann, S., Aimone, C.: OpenVIDIA: Parallel GPU computer vision. In: Proc. of ACM Multimedia, pp. 849–852 (2005)Google Scholar
  7. 7.
    Fung, J., Mann, S.: Using graphics devices in reverse: GPU-based image processing and computer vision. In: Proc. IEEE Int. Conf. Multimedia Expo., pp. 9–12 (2008)Google Scholar
  8. 8.
    Govindaraju, N.K.: GPUFFTW: High performance GPU-based FFT library. In: Supercomputing (2006)Google Scholar
  9. 9.
    Grauer-Gray, S., Kambhamettu, C., Palaniappan, K.: GPU implementation of belief propagation using CUDA for cloud tracking and reconstruction. In: Proc. PRRS, pp. 1–4 (2008)Google Scholar
  10. 10.
    Guan, S., Klette, R., Woo, Y.W.: Belief propagation for stereo analysis of night-vision sequences. In: Wada, T., Huang, F., Lin, S. (eds.) PSIVT 2009. LNCS, vol. 5414, pp. 932–943. Springer, Heidelberg (2009)CrossRefGoogle Scholar
  11. 11.
    Klette, R.: Analysis of data flow for SIMD systems. Acta Cybernetica 6, 389–423 (1984)Google Scholar
  12. 12.
    NVIDIA. NVIDIA CUDA Programming Guide Version 2.1 (2008),
  13. 13.
    Scharstein, D., Szeliski, R.: A taxonomy and evaluation of dense two-frame stereo correspondence algorithms. Int. J. Computer Vision 47, 7–42 (2002)zbMATHCrossRefGoogle Scholar
  14. 14.
    Sinha, S.N., Frahm, J.M., Pollefeys, M., Genc, Y.: Feature tracking and matching in video using graphics hardware. In: Proc. Machine Vision and Applications (2006)Google Scholar
  15. 15.
    Vineet, V., Narayanan, P.J.: CUDA cuts: Fast graph cuts on the GPU. In: CVPR Workshop on Visual Computer Vision on GPUs (2008)Google Scholar
  16. 16.
    Vaudrey, T., Klette, R.: Residual Images Remove Illumination Artifacts for Correspondence Algorithms! In: Proc. DAGM (to appear, 2009)Google Scholar
  17. 17.
    Yang, Q., Wang, L., Yang, R., Wang, S., Liao, M., Nistér, D.: Real-time global stereo matching using hierarchical belief propagation. In: Proc. British Machine Vision Conf., pp. 989–998 (2006)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Yanyan Xu
    • 1
  • Hui Chen
    • 1
  • Reinhard Klette
    • 2
  • Jiaju Liu
    • 1
  • Tobi Vaudrey
    • 2
  1. 1.School of Information Science and EngineeringShandong UniversityChina
  2. 2.The .enpeda.. ProjectThe University of AucklandNew Zealand

Personalised recommendations