7.1 Conclusions
In this thesis, we design two algorithms for different goals. Thus, we conclude them in two parts.
In first part from Chapter 3 to Chapter 5, we study the computational complexity of building CU quadtree. Our fast CU size decision algorithm, which is based on the size information of the neighboring CUs and the co-located CU, speeds up the encoding procedure at about 1.75 times faster in average comparing to the original encoding process. Then, we also combine the existing ECU and CFM schemes together with our proposed algorithm in an efficient way. In the low QP cases, our algorithm provides more time reduction over ECU and CFM with acceptable coding loss. Totally, the combined fast algorithm offers averagely 51%
time reduction, and the BD-rate increases at about 2.02% for high resolution videos. Our algorithm is also particularly useful in the low motion videos such as vidyo1, vidyo3, and vidyo4. This type of videos often occur in the mobile video communication, and the combined algorithm achieves up to about 69% time reduction with tolerable BD-PSNR drop about 0.05dB for the test sequences.
Chapter 6 is the second part: We study the effect of transform on motion vector selection.
We propose the modified AVC/H.264 motion vector search process. First, we keep five
integer MV candidates by their SAD values, and then process the sub-pixel MV searching by using SATD. At the end, we use the modified AVC mode decision function to estimate the R-D cost to decide the best MV from five candidates. In comparing to the previous approach, our method not only reduces the time-consuming SATD calculations but also avoids the poor performance of using SATD directly in integer MV selection. In general, our proposed optimization scheme achieves 2.3% bit rate saving with an additional 45% encoding time, averagely. The method can achieve up to 4.2% BD-rate improvement in our test sequences, and the algorithm performs well especially for the sequences with strong residual texture.
7.2 Future Work
In proposing fast algorithms for HEVC, we design our algorithm under the configuration of low complexity and low delay P, but we change the encoding parameter setting in the GOP size and the number of reference frames. For real applications, we should consider the incremented QP to adjust the decision rule and the thresholds, adaptively. On the other hand, considering the MV offset and the multiple reference frames in the search of co-located CU will decrease the coding loss for our proposed fast algorithm. Last but not the least, we can include other indicators in reducing candidates. For example, cbf is an important indicator telling us whether the nearby CU partitions are reliable or not, especially for the termination decision in large size CU. Reducing the coding loss in the low bitrate case is a research
challenge.
Another topic is the R-D performance improvement for H.264/AVC. Its bottleneck is the high complexity. Therefore, we suggest some methods for speeding up the modified encoding procedure at the end of section 6.3. If we want to extend this combined ME and transform idea to HEVC, the scheme will be very complicated because HEVC has transform of different sizes. Also, because the current HEVC has very flexible ME modes and transform modes, the combined scheme may not provide much additional advantage.
REFERENCES
[1] T. Wiegand et al., “Draft ITU-T Recommendation and Final Draft International Standard of Joint Video Specification (ITU-T Rec. H.264 | ISO/IEC 14496-10 AVC),”
ISO/IEC JTC/SC29/WG11 and ITU-T SG16 Q.6, JVT-Go50r1, Mar.2003.
[2] H. M. Hang et al., “Towards the next video standard: high efficiency video coding,”
Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, 2010.
[3] B. Girod, “The Efficiency of Motion-compensating prediction for hybrid coding of video sequences,” IEEE Journal on Selected Areas in Communications, vol. 5, no. 7, pp.
1140-1154, 1987.
[4] I. E. G. Richardson, H.264 and MPEG-4 Video Compression: Video Coding for Next-Generation Multimedia, Wiley.
[5] T. Wiegand, et. al., “Rate-constrained coder control and comparison of video coding standards,” IEEE Trans. Circuit Syst. Video Technol., vol. 13, no. 7, pp. 688–703, Jul.
2003.
[6] T.Wiegand and B. Girod, “Lagrange multiplier selection in hybrid video coder control,”
in Proc. Int. Conf. Image Proc., pp542–545, Oct. 2001.
[7] T. Wiegand et al., “Special section on the joint call for proposals on High Efficiency
20, no. 12, pp. 1661–1666, 2010.
[8] D. Marpe et al., “Video compression using nested quadtree structures, leaf merging, and improved techniques for motion representation and entropy coding,” IEEE Trans.
Circuits Syst. Video Technol., vol. 20, no. 12, pp. 1676–1687, 2010.
[9] JCT-VC, “High Efficiency Video Coding (HEVC) Test Model 5 (HM 5) Encoder Description”, JCT-VC document, JCTVC-G1102, January 2012.
[10] P. A. Chou et al., “Optimal pruning with applications to tree-structured source coding and modeling,” IEEE Trans. Inform. Theory, vol. 35, no. 2, pp. 299–315, Mar. 1989.
[11] G. J. Sullivan and R. L. Baker, “Efficient quadtree coding of images and video,” IEEE Trans. Image Process., vol. 3, no. 3, pp. 327–331, May 1994.
[12] JCT-VC, “WD5: Working Draft 5 of High-Efficiency Video Coding”, JCT-VC document, JCTVC-G1103, December 2012.
[13] H.264/AVC codec [online] http://iphome.hhi.de/suehring/tml/download/.
[14] Test sequences of H.264/AVC [Online] http://media.xiph.org/video/derf/ . [15] HEVC codec [online] http://hevc.kw.bbc.co.uk/trac/browser/tags/HM-5.0 .
[16] G. Bjontegaard, “Calculation of Average PSNR Differences between RD-curves,” Document VCEG-M33, Apr. 2001.
[17] J. Kim et al., “Adaptive coding unit early termination algorithm for HEVC,”
International Conference on Consumer Electronics, pp. 261–262, 2012.
[18] K. Choi et al., “Coding tree pruning based CU early termination,” JCT-VC document, JCTVC-F092, Jul. 2011.
[19] R. H. Gweon et al., “Early termination of CU encoding to reduce HEVC complexity,”
JCT-VC document, JCTVC-F045, Jul. 2011.
[20] J. Leng et al., “Content based hierarchical fast coding unit decision algorithm for HEVC,” International Conference on Multimedia and Signal Processing, pp. 56–59, 2011.
[21] G. Correa and L. Agostini, “Complexity control of high efficiency video encoders for power-constrained devices,” IEEE Trans. on Consumer Electronics, Vol. 57, No. 4, Nov.
2011.
[22] H. S. Malvar et al., “Low complexity transform and quantization in H.264/AVC,” IEEE Trans. Circuit Syst. Video Technol., vol. 13, no. 7, pp. 598–603, Jul. 2003.
[23] A. Abdelazim et al., “Effect of the hadamard transform on motion estimation of different layers in video coding,” International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XXXVIII, Part5 Commission V Symposium, Newcastle upon Tyne, UK. 2010.