為因應未來高解析度及影像串流需求，現行的影像編碼標準H.264/MPEG4 Advanced Video Coding (AVC) 已不敷使用，JCT-VC (Joint Collaborative Team on Video Coding) 在2013年發表下一代編碼標準High Efficiency Video Coding (HEVC)，HEVC提供相較於H.264/MPEG4 AVC更多不同大小的編碼區塊，有更好的壓縮比，但大為提升的計算複雜度。本論文專注在使用GPU來加速HEVC編碼中的動態估計及PU模式選擇。實驗結果顯示，平均而言，可以在-0.001%~0.194%的PSNR損失及3.106%~9.906%的bit-rate增加下，達到15.12~17.62倍的編碼效能提升。

For the request of high resolution and video streaming, the H.264/MPEG4 Advanced Video Coding, aka H.264/MPEG4 AVC, is not adequate for needs. In 2013, Joint Collaborative Team on Video Coding (JCT-VC) published the next generation video coding standard, as called High Efficiency Video Coding (HEVC). HEVC supports more different size of coding block than H.264/MPEG4 AVC, and it leads better video compression ratio and high computation complexity. In the thesis, we focus on using GPU to accelerate the motion estimation and PU mode decision of HEVC encoding. In the experiment, our work can achieve the average 15.12~17.62 times speed-up with 3.106%~9.906% bit-rate increase and -0.001%~0.194% PSNR loss.

Chinese Abstract i
Abstract ii
Contents iv
List of Figures vi
List of Tables vii
1 Introduction 1
2 Background 3
2.1 Concepts of HEVC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3 Methodology 11
3.1 Proposed method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4 Implementation 14
4.1 The parallelism strategy in x265 . . . . . . . . . . . . . . . . . . . . . 14
4.2 Working concurrently between CPU and GPU . . . . . . . . . . . . . 14
4.3 The performance issues on GPU . . . . . . . . . . . . . . . . . . . . . 15
4.4 Implementation of proposed method . . . . . . . . . . . . . . . . . . 16
5 Experiment 20
6 Conclusion and Future Work 30
References 31
A Block-matching algorithm 33
B The results of experiment 35

[1] x265. https://bitbucket.org/multicoreware/x265/. Accessed: 2015-07-09.
[2] Cuda c programming guide. http://docs.nvidia.com/cuda/
cuda-c-programming-guide/, 2015-03-05. Accessed: 2015-07-09.
[3] Jiang, X., Song, T., Shimamoto, T., and Wang, L. High efficiency video coding (hevc) motion estimation parallel algorithms on gpu. In Consumer Electronics - Taiwan (ICCE-TW), 2014 IEEE International Conference on (May 2014), pp. 115-116.
[4] Ma, J., Luo, F., Wang, S., and Ma, S. Flexible ctu-level parallel motion estimation by cpu and gpu pipeline for hevc. In Visual Communications and Image Processing Conference, 2014 IEEE (Dec 2014), pp. 282-285.
[5] Monteiro, E., Maule, M., Sampaio, F., Diniz, C., Zatt, B., and Bampi, S. Real-time block matching motion estimation onto gpgpu. In Image Processing (ICIP), 2012 19th IEEE International Conference on (Sept 2012), pp. 1693-1696.
[6] Radicke, S., Hahn, J., Grecos, C., and Wang, Q. A highly-parallel approach on motion estimation for high efficiency video coding (hevc). In Consumer Electronics (ICCE), 2014 IEEE International Conference on (Jan 2014), pp. 187-188.
[7] Sullivan, G., Ohm, J., Han, W.-J., and Wiegand, T. Overview of the high efficiency video coding (hevc) standard. Circuits and Systems for Video Technology, IEEE Transactions on 22, 12 (Dec 2012), 1649-1668.
[8] Wang, F., Zhou, D., and Goto, S. Opencl based high-quality hevc motion estimation on gpu. In Image Processing (ICIP), 2014 IEEE International Conference on (Oct 2014), pp. 1263-1267.
[9] wen Wang, X., Song, L., Chen, M., and jie Yang, J. Paralleling variable block size motion estimation of hevc on multi-core cpu plus gpu platform. In Image Processing (ICIP), 2013 20th IEEE International Conference on (Sept 2013), pp. 1836-1839.
[10] Zhu, S., and Ma, K.-K. A new diamond search algorithm for fast block matching
motion estimation. Image Processing, IEEE Transactions on 9, 2 (Feb 2000), 287-290.