Parallel Implementations of the CSBP Stereo Vision Algorithm
Resumo
We developed two parallel versions of the Constant Space Belief Propagation algorithm (CSBP - one of the best stereo algorithms currently known) [1]: one in OpenMP and one in C+CUDA. For images with 640x480 pixels, the sequential version has a performance of 1.16 frames per second (FPS), the OpenMP parallel version has a performance of 3.7 FPS, while the C+CUDA version has a performance of 17.3 FPS in high-performance desktop machines. These results are important because they enable the implementation of autonomous vehicles with sensors like camera, which is one of the objectives of a PRONEX project being currently developed in LCADDI/ UFES. One of the goals of this project is to implement an autonomous vehicle from a commercial automobile.
Referências
S. Thrun, W. Burgard, D. Fox. “Probabilistic Robotics”, MIT Press, 2005.
C.A. Carvalho, L.P. Veronessi, H. Oliveira, A.F. De Souza. “Implementation of a Biologically Inspired Stereoscopic Vision Model in C+CUDA”, accepted in GPU Tecnology Conference, Sep 30, 2009.
Middlebury Stereo Vision Page – http://vision.middlebury.edu/stereo/, consultado em 05/08/2011
W.T. Freeman, E. Pasztor, and O.T. Carmichael. “Learning low-level vision”, IJCV, 40(1):25–47, 2000.
J. Sun, N. Zheng, and H.Y. Shum. “Stereo matching using belief propagation”, PAMI, 25(7):787–800, 2003.
Q. Yang, C. Engels, and A. Akbarzadeh. “Near realtime stereo for weakly-textured scenes”, In BMVC, pages 80–87, 2008
T. Yu, R.S. Lin, B. S., B. Tang. “Efficient message representations for belief propagation”. In ICCV , pages 1–8, 2007.
B. Chapman, G. Jost, R.V. der Pas, David J. Kuck. “Using OpenMP: Portable Shared Memory Parallel Programming”. Scientific and Engineering Computation Series, The MIT Press. 2008.
Point Grey – Stereo Vision – Bumblebee2 CCD FireWire camera , http://www.ptgrey.com, consultado em 05/08/2011.
NVIDIA, “NVIDIA’s Next Generation CUDA Compute Architecture: Fermi”, 2009
NVIDIA, “NVIDIA CUDA: Compute Unified Device Architecture - Programming Guide 4.0”, 2011.
NVIDIA, “Technical Brief: NVIDIA GeForce GTX200 GPU Architectural Overview”, 2008.
