臺灣博碩士論文加值系統

近幾年來，隨著多媒體通訊產品逐漸成長，使用者對視訊的品質和傳輸速度也愈來愈要求，不但要求傳輸速度快，而且視訊的品質也需要有相當的水準，在最近幾年來，H.263+已發展成一個很成熟的壓縮標準，已應用於多種產品上，但隨著多媒體傳輸的資料量愈來愈大，H.263+的壓縮效能已然不足。
而最新所提出來的H.264壓縮標準，採用更複雜的編碼方式，包括可使用較小的區塊來做預測，亦可往前參考五張畫面，在非整數點搜尋方面更可搜尋到四分之一點的精確度，使預測更加精準，提供比H.263+更低的位元率和更好視訊品質，但因使用更多不同的模式做預測以及往前參考多張畫面，所以增加了動量預測的複雜度，雖可提供很好的效能，但其編碼速度過慢為一大缺點，而隨著硬體快速地發展，硬體的執行速度相當地快速，但目前的硬體設備仍不足使H.264編碼速度達到即時編碼的目的。
所以在本論文中，提出了一種快速而有效率的動量預測的方式，針對在搜尋多張參考的畫面時，重覆利用前一張參考畫面的向量資訊，組合出欲搜尋的向量，大幅減少搜尋時間，而結果也能近似於全域搜尋的效果，在位元率和主觀與客觀性效果近似的條件下，提供了四至五倍的編碼速度，期望能達到即時編碼的目的。

In recent years, H.263+ has been developed to a mature stage for video-conferencing and many other applications. Due to the development rapid of hardware and the Internet, the new H.264 standard will enhance the H.263’s capability and performance, and it is expected to be the next generation standard and achieve real-time performance.
In the H.264 standard, a lot of time will be consumed in an encoder for motion estimation. It uses seven block sizes to perform the motion compensation, and refer previous five frames for the searching of the best motion vector to achieve lower bit-rate and higher PSNR. And the H.264 reference software uses a full search scheme to obtain the best performance. Thus, it will spend a lot of searching time.
In this thesis, we propose an efficient searching algorithm which reuses the motion vector information from the last reference frame. The proposed algorithms use the stored motion vectors to synthesize the current motion vector without performing the motion estimation in each reference frame. Therefore, our proposed algorithms can obtain the speed up ratio by 4 to 5 times for encoding, which benefits from the prediction of the reference frames in advance and maintain the good performance.

List of Content
Chapter 1 Introduction8
1.1 Overview of Video Coding8
1.2 Motivation18
1.3 The organization of the thesis19
Chapter 2 Motion Estimation in H.264 Standard20
2.1 Overview of the H.264 Standard21
2.1.1 Source Coder21
2.1.2 Motion Estimation and Mode Decision23
2.2 Review of previous works38
2.2.1 Analysis and Reduction of Multi-Reference Frames38
2.2.2 A Novel Approach to Fast Multi-Frame Selection for Fractional pixel accuracy43
Chapter 3 The Proposed Efficient Multi-Frame Motion Estimation Algorithms48
3.1 Re-use of the motion vectors in each reference frame49
3.2 Forward Dominant Vector Selection (FDVS) for multi-frame motion estimation50
3.2.1 FDVS for multi-frame motion estimation in H.264 reference software51
3.3 Variable block size Activity Dominant Vector Selection (VADVS)55
3.3.1 Activity Dominant Vector Selection (ADVS)55
3.3.2 The proposed Variable block size Activity Dominant Vector Selection (VADVS)56
3.3.3 Adaptive VADVS algorithm (AVADVS)62
3.4 The refinement for the composed motion vector68
Chapter 4 Experiment Result and Performance Analysis70
4.1 Parameter setting71
4.2 Rate distortion curve comparison71
4.3 Variation of bit-rate comparison75
4.4 PSNR Comparison80
4.5 Speed up ratio comparison84
4.6 Comparison of bit-rate, PSNR and speed up ratio88
4.7 MV difference comparison91
4.8 Memory analysis95
4.9 Subjective quality comparison96
4.9.1 Low Bit-Rate (9k bps)97
4.9.2 Median Bit-Rate (24k bps)99
4.9.3 High Bit-Rate (64k bps)101
4.10 Performance Analysis104
Chapter 5 Conclusion and Future Work105
Bibliography107

Bibliography
[1]Video Coding for Low Bit-rate Communication, ITU-T Recommandation H.263 Version 2 (H.263+), January 1998.
[2]Joint Video Team (JVT) of ISO/IEC MPEG and ITU-T VCEG, “Working Draft Number 2, Revision 2,” Geneva, Switzerland, January 29-February 1, 2002.
[3]Joint Video Team (JVT) of ISO/IEC MPEG and ITU-T VCEG, “Draft ITU-T Recommendation and Final Draft International Standard of Joint Video Specification (ITU-T Rec. H.264 | ISO/IEC 14496-10 AVC),” document JVT-G050d35.doc, March 2003.
[4]International Telecommunication Union, “Video codec for audio visual services at p*64 Kbps,” ITU-T Recommendation H.261, March 1993.
[5]ISO/IEC JTC1/SC29/WG11, “ISO/IEC CD 11172: Information Technology,” MPEG-2 Committee Draft, December 1991.
[6]Joint Video Team (JVT) software JM61E, March 2003.
[7]Yu-Wen Huang, Bing-Yu Hsieh, Tu-Chih Wang, Shao-K Chien, Shyh-Yih Ma, Chun-Fu Shen and Liang-Gee Chen, “Analysis and reduction of reference frames for motion estimation in MPEG-4 AVC/JVT/H.264,” IEEE International Conference on Acoustics, Speech, and Signal Processing, Vol. 3, pp. III_145 -III_148, April 6-10, 2003.
[8]A. Chang, O.C. Au, Y.M. Yeung, “A novel approach to fast multi-frame selection for H.264 video coding,” International Symposium on Circuits and Systems Proceedings, Vol. 2, pp. 704 -707, 2003.
[9]Mei-Juan Chen, Ming-Chung Chu and Chih-Wei Pan, “Efficient motion estimation algorithm for reduced frame-rate video transcoder,” IEEE Transactions on Circuits and Systems for Video Technology, Vol.12, No.4, pp. 269 -275, 2002.
[10]Jeongnam Youn and Ming-Ting Sun, “A fast motion vector composition method for temporal transcoding,” IEEE International Symposium on Circuits and Systems Proceedings. Vol.4, pp. 243 -246, 1999.
[11]Joint Video Team software JM6.1E, March 2003.
http://bs.hhi.de/~suehring/tml/download/