Fast and Efficient Dense Variational Stereo on GPU

Thanks to their high performance and programmability, the latest graphics cards can now be used for scientific purpose. They are indeed very efficient parallel single instruction multiple data (SIMD) machines. This new trend is called general purpose computation on graphics processing unit (GPGPU). Regarding the stereo problem, variational methods based on deformable models provide dense, smooth and accurate results. Nevertheless, they prove to be slower than usual disparity-based approaches. In this paper, we present a dense stereo algorithm, handling occlusions, using three cameras as inputs and entirely implemented on a graphics processing unit (GPU). Experimental speedups prove that our approach is efficient and perfectly adapted to the GPU, leading to nearly video frame rate reconstruction.

[1]  O. Faugeras,et al.  Variational principles, surface evolution, PDE's, level set methods and the stereo problem , 1998, 5th IEEE EMBS International Summer School on Biomedical Imaging, 2002..

[2]  Julien Mairal,et al.  A GPU implementation of variational stereo , 2006 .

[3]  Markus Hadwiger,et al.  Accurate Dense Stereo Reconstruction using Graphics Hardware , 2003, Eurographics.

[4]  Reinhard Koch,et al.  Multi-camera real-time depth estimation with discontinuity handling on PC graphics hardware , 2004, Proceedings of the 17th International Conference on Pattern Recognition, 2004. ICPR 2004..

[5]  D. Scharstein,et al.  A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms , 2001, Proceedings IEEE Workshop on Stereo and Multi-Baseline Vision (SMBV 2001).

[6]  Ruigang Yang,et al.  A Unified Approach To Real-Time, Multi-Resolution, Multi-Baseline 2d View Synthesis And 3d Depth Estimation Using Commodity Graphics Hardware , 2004, Int. J. Image Graph..

[7]  Luc Van Gool,et al.  Fast interpolated cameras by combining a GPU based plane sweep with a max-flow regularisation algorithm , 2004, Proceedings. 2nd International Symposium on 3D Data Processing, Visualization and Transmission, 2004. 3DPVT 2004..

[8]  Minglun Gong,et al.  Near real-time reliable stereo matching using programmable graphics hardware , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[9]  L. Van Gool,et al.  Fast interpolated cameras by combining a GPU based plane sweep with a max-flow regularisation algorithm , 2004 .

[10]  Horst Bischof,et al.  Hierarchical Disparity Estimation with Programmable 3D Hardware , 2004 .

[11]  Reinhard Koch,et al.  Real-time multi-stereo depth estimation on GPU with approximative discontinuity handling , 2004 .

[12]  Randima Fernando,et al.  The GeForce 6 series GPU architecture , 2005, SIGGRAPH Courses.

[13]  Olivier D. Faugeras,et al.  Variational principles, surface evolution, PDEs, level set methods, and the stereo problem , 1998, IEEE Trans. Image Process..

[14]  Ruigang Yang,et al.  A versatile stereo implementation on commodity graphics hardware , 2005, Real Time Imaging.