Performance engineering to achieve real-time high dynamic range imaging
详细信息 查看全文
- 作者：Harald Köstler ; Markus Stürmer ; Thomas Pohl
- 关键词：High dynamic range imaging ; GPGPU ; Multigrid ; Performance model
- 刊名：Journal of Real-Time Image Processing
- 出版年：2016
- 出版时间：January 2016
- 年：2016
- 卷：11
- 期：1
- 页码：127-139
- 全文大小：1,173 KB
- 参考文献：1.Brandt, A.: Multi-level adaptive solutions to boundary-value problems. Math. Comput. 31(138), 333–390 (1977)MATH MathSciNet CrossRef
2.Briggs, W., Henson, V., McCormick, S.: A Multigrid Tutorial, 2nd edn. Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA, USA (2000)
3.Christadler, I., Köstler, H., Rüde, U.: Robust and efficient multigrid techniques for the optical flow problem using different regularizers. In: Hülsemann F, Kowarschik M, Rüde U (eds) Proceedings of 18th Symposium Simulationstechnique ASIM 2005, SCS Publishing House, Erlangen, Germany, Frontiers in Simulation, vol. 15, pp. 341–346, preprint version published as Tech. Rep. 05-6 (2005)
4.Christen, M., Schenk, O., Burkhart, H.: Patus: A code generation and autotuning framework for parallel iterative stencil computations on modern microarchitectures. In: In International Parallel and Distributed Processing Symposium (IPDPS), IEEE, pp. 676–687 (2011)
5.Douglas C., Hu J., Kowarschik M., Rüde U., Wei C.: Cache optimization for structured and unstructured grid multigrid. Electr. Transact. Numer. Anal. (ETNA) 10:21–40 (2000)MATH
6.Fattal R., Lischinski D., Werman M.: (2002) Gradient domain high dynamic range compression. ACM Transact. Graph. 21(3):249–256CrossRef
7.Frigo, M., Strumpen, V.: Cache oblivious stencil computations. In: Arvind, Rudolph L (eds.) Proceedings of the 19th annual international conference on Supercomputing, ACM, pp 361–366 (2005)
8.Gradl, T., Freundl, C., Köstler, H., Rüde, U.: Scalable multigrid. In: Wagner, S., Steinmetz, M., Bode, A., Brehm, M. (eds) High Performance Computing in Science and Engineering. Garching/Munich 2007, LRZ, KONWIHR, Springer, Berlin, pp. 475–483 (2008)
9.Hackbusch, W.: Multi-Grid Methods and Applications. Springer, Berlin (1985)
10.He, K., Sun, J., Tang, X.: Guided image filtering. Computer Vision–ECCV 2010 pp. 1–14 (2010)
11.Hülsemann, F., Kowarschik, M., Mohr, M., Rüde, U.: Parallel geometric multigrid. In: Bruaset A, Tveito A (eds) Numerical Solution of Partial Differential Equations on Parallel Computers, Lecture Notes in Computational Science and Engineering, vol. 51, Springer, Berlin, chap. 5, pp. 165–208 (2005)
12.Kamil, S., Chan, C., Oliker, L., Shalf, J., Williams, S.: An auto-tuning framework for parallel multicore stencil computations. In: In International Parallel and Distributed Processing Symposium (IPDPS), pp. 1–12 (2010)
13.Köstler, H.: A Multigrid Framework for Variational Approaches in Medical Image Processing and Computer Vision. PhD thesis, Friedrich-Alexander-Universität Erlangen-Nürnberg (2008)
14.Köstler H., Stürmer M., Rüde U.: A fast full multigrid solver for applications in image processing. Numerical linear algebra with applications 15(2–3):187–200 (2008)MATH MathSciNet
15.Kowarschik, M., Wei, C., Rüde, U.: DiMEPACK: a cache–optimized multigrid library. In: Arabnia H (ed.) Proceedings of the International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA 2001), CSREA Press, Irvine, CA, USA, Las Vegas, NV, USA, vol. I, pp. 425–430 (2001)
16.Levin, A., Zomet, A., Peleg, S., Weiss, Y.: Seamless image stitching in the gradient domain. Lecture Notes in Computer Science, Computer Vision-ECCV 2004 3024/2004:377–389 (2004)
17.McCann, J., Pollard, N.: Real-time gradient-domain painting. ACM Transact. Graphics-TOG 27(3):93–93 (2008)
18.Schäfer, A., Fey, D.: High Performance Stencil Code Algorithms for GPGPUs. Procedia Comput. Sci. 4, 2027–2036 (2011)CrossRef
19.Stürmer, M., Treibig, J., Rüde, U.: Optimising a 3d multigrid algorithm for the IA-64 architecture. Int. J. Comput. Sci. Eng. 4(1), 29–35 (2008)
20.Stürmer, M., Wellein, G., Hager, G., Köstler, H., Rüde, U.: Challenges and potentials of emerging multicore architectures. In: Wagner, S., Steinmetz, M., Bode, A., Brehm, M. (eds.) High Performance Computing in Science and Engineering. Garching/Munich 2007, LRZ, KONWIHR, Springer, Berlin, pp. 551–566 (2008)
21.Trottenberg, U., Oosterlee, C., Schüller, A.: Multigrid. Academic Press, San Diego, CA, USA (2001)
22.Wang, T., Ke, W., Zwao, D., Chen, F., Chiu, C: Block-based gradient domain high dynamic range compression design for real-time applications. In: Image Processing, 2007. ICIP 2007. IEEE International Conference on, IEEE Computer Society Washington, DC, USA, vol. 3, pp. 561–564 (2007)
23.Wienands, R., Joppich, W.: Practical Fourier analysis for multigrid methods, Numerical Insights, vol. 5, Chapmann and Hall/CRC Press, Boca Raton, Florida, USA (2005) - 作者单位：Harald Köstler (1)
Markus Stürmer (1)
Thomas Pohl (2)
1. Universität Erlangen-Nürnberg, Lehrstuhl für Systemsimulation, Cauerstr. 11, 91058, Erlangen, Germany
2. Siemens AG, Healthcare Sector, Angiography and Interventional X-Ray Systems, Siemenstr. 1, 91301, Forchheim, Germany - 刊物类别：Computer Science
- 刊物主题：Image Processing and Computer Vision
Multimedia Information Systems
Computer Graphics
Pattern Recognition
Signal,Image and Speech Processing - 出版者：Springer Berlin / Heidelberg
- ISSN：1861-8219
文摘
Image-processing applications like high dynamic range imaging can be done efficiently in the gradient space. For it, the image has to be transformed to gradient space and back. While the forward transformation to gradient space is fast by using simple finite differences, the backward transformation requires the solution of a partial differential equation. Although one can use an efficient multigrid solver for the backward transformation, it shows that a straightforward implementation of the standard algorithm does not lead to satisfactory runtime results for real-time high dynamic range compression of larger 2D X-ray images even on GPUs. Therefore, we do a rigorous performance analysis and derive a performance model for our multigrid algorithm that guides us to an improved implementation, where we achieve an overall performance of more than 25 frames per second for 16.8 Megapixel images doing full high dynamic range compression including data transfers between CPU and GPU. Together with a simple OpenGL visualization it becomes possible to perform real-time parameter studies on medical data sets.