簡易檢索 / 詳目顯示

研究生: 林義席
Lin, Yi-Hsuv
論文名稱: 取代H.264 內部編碼之位元率-失真模組的低複雜度演算法
A Complexity Reduced Algorithm for H.264 Intra Rate & Distortion Module
指導教授: 戴顯權
Tai, Shen-Chuan
學位類別: 碩士
Master
系所名稱: 電機資訊學院 - 電腦與通信工程研究所
Institute of Computer & Communication Engineering
論文出版年: 2005
畢業學年度: 93
語文別: 英文
論文頁數: 45
中文關鍵詞: 內部預測模式快速模式選擇
外文關鍵詞: fast intra prediction mode decision, fast intra RDO
相關次數: 點閱:80下載:1
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  • 一般在H.264 的內部編碼(Intra coder)中有2種常用來選擇預測模式的方法,分別是絕對誤差總和(SAD)跟位元率-失真最佳化模組(Rate & Distortion Optimization) ,這兩種運算擁有各自的好處,SAD 是快速而簡單的方法,其編碼效果較為粗糙,相反地,RDO 決定預測模式的方法則是以增加計算時間的方式來算出兼具高影像品質跟壓縮率的預測模式,因此其效能表現非常良好,但是它同時也具有運算時間太長的問題。
    本論文提出一個能夠兼顧雙方優點的演算法,試著提升品質及壓縮率雙方面的效能,並同時保有低計算時間的特性。演算法的主體分成2個階段,第一階段時我們會先針對每一個4x4內部方塊自訂一種取樣樣板,並對每一個預測模式只取出這些樣板中的SAD值,接著利用分析得到的SAD 門檻值,把擁有過大SAD 的預測模式濾掉,保留幾個預測良好的候選模式,接著把這些候選模式放入第二階段的測試,在第二階段我們利用一個類似RDO 的流程以取得影像品質跟壓縮率的平衡,在輸出位元率的資訊時,不同於原本RDO法真實地去計算位元率,我們提出一個快速地估計位元率的方式來取得位元率資訊。
    實驗數據顯示我們的演算法能適用於各種複雜度的影像,並有效的將效能提升到接近RDO的水準,而計算時間相對於RDO,能節省的比率介於94.51%到89.03%之間。

    There are two common methods to select the prediction mode in H.264 intra coder, and they are SAD method and Rate & Distortion Optimization (RDO) module respectively. They both have their own advantages and disadvantages. SAD method is a fast and simple method to select prediction mode, but its performance is a little bit coarse. On the contrary, RDO increases a lot of time to select the prediction mode, which is a balance point between the compression ratio and high image quality, so RDO performs a very good performance, but it also exists a problem of high computation time.
    This Thesis shows an algorithm which can combine the advantages of these two mentioned methods, and try to increase the performance both in compression ratio and image quality with the property of low computation time. The main structure of the proposed algorithm is a Two-level mode decision. First we define some sampled points of every 4x4 intra block, then only calculate the SAD value for the sampled points at Level-one. According to the thresholds analyzed in the Thesis, we filter out some prediction modes which have too large SAD, and the other modes are retained as candidate modes. Candidate modes are inputted into Level-two test. The Level-two test adapts the method like RDO to balance the compression ratio and the image quality. When outputting bit rate, it is different from RDO which compute the actual bit rate. We develop a fast method to approximate bit rate instead of the actual bit rate calculation.
    Experimental results shows that our algorithm could be adaptive for any videos with different type of complexity, and effectively improve the performance near by RDO then compared with RDO the proposed algorithm could save the computation time from 94.51% to 89.03%.

    LIST OF TABLES................................................................ii LIST OF FIGURES..............................................................iii CHAPTER 1 Introduction.........................................................1   1.1 Compression of Image Coding Systems ...................................2 CHAPTER 2 Background...........................................................6   2.1 Sum of Absolute Difference Method......................................8   2.2 Rate & Distortion Optimization (RDO)..................................10 CHAPTER 3 The Proposed Two-Level Intra Mode Decision Algorithm................13   3.1 Level-one: Coarse Mode Decision ......................................14     3.1.1 Compute the sampled SAD.........................................15     3.1.2 Early Termination...............................................19     3.1.3 Select Candidate Mode...........................................20   3.2 Level-two: Refinement Pass............................................22     3.2.1 Distortion..................................................... 24     3.2.2 Approximation Function for Bit Rate.............................24     3.2.3 Modified Lagrangian Function....................................28 CHAPTER 4 Experimental Results................................................29 CHAPTER 5 Conclusions ........................................................42 Reference ....................................................................43 Biography ....................................................................45

    [1] Draft ITU-T Recommendation and Final Draft International Standard of Joint Video Specification, May 2003.
    [2] Information Technology—Coding of Audio-Visual Objects—Part 2: Visual, 1999.
    [3] Video Coding for Low Bit Rate Communication, 1998.
    [4] Information Technology—Generic Coding of Moving Pictures and Associated Audio Information: Video, 1996.
    [5] A. Joch, F. Kossentini, H. Schwarz, T.Wiegand, and G. J. Sullivan, “Performance comparison of video coding standards using lagragian coder control,” in Proc. IEEE Int. Conf. Image Processing, 2002, pp. 501–504.
    [6] Information Technology—Digital Compression and Coding of Continuous- Tone Still Images, 1994.
    [7] JPEG 2000 Part I, Mar. 2000. ISO/IEC JTC1/SC29/WG1 Final Committee Draft, Rev. 1.0.
    [8] T.Wiegand, G. J. Sullivan and G. Bjontegaard and A. Luthra, ” verview of the H.264/AVC Video Coding Standard,” IEEE Transactions on Circuits, System and Video Technology, Vol. 7, pp. 1-19, July 2003.
    [9] F. Pan, X. Lin, et. al., ”Fast Mode Decision for Intra Prediction,” ISO/IEC JTC1/SC29/WG11 and ITU-T SG16 Q.6, JVT 7th Meeting Pattaya II, Thailand, March 2003.
    [10] Changsung Kim, Qing Li and C. C. Jay Kuo, ”Fast Intra-prediction model selection for H.264 codec,” SPIE International Symposium ITCOM 2003, Orlando, Florida, July, 2003.
    [11] Changsung Kim, Hsuan-Huei Shih and C. C. Jay Kuo, ”Multistage Mode 44 Decision for Intra Prediction in H.264 Codec,” IS&T/SPIE 16th Annual Symposium EI, Visual Communications and Image Processing, Orlando, Florida, January, 2004.
    [12] Changsung Kim, Hsuan-Huei Shih and C. C. Jay Kuo, ”Feature-Based Intra-Prediction Mode Decision for H.264”, IEEE Proceedings of International Conference Image Processing, submitted, Singapole, October, 2004.
    [13] Zhihai He and Sanjit K. Mitra, ”A Unified Rate Distortion Analysis Framework for Transform Coding,” IEEE Transactions on Circuits, System and Video Technology, Vol. 11, No. 12, Dec 2001.

    下載圖示 校內:2006-08-04公開
    校外:2006-08-04公開
    QR CODE