請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/40665
標題: | 強健性區塊模式選擇演算法之設計與實作 Design and Implementation of Robust Mode Decision Algorithms for Video Transmission over Lossy Networks |
作者: | Ya-Lun Yang 楊雅倫 |
指導教授: | 謝宏昀 |
關鍵字: | 視訊編碼,區塊模式, Video coding,Mode selection, |
出版年 : | 2011 |
學位: | 碩士 |
摘要: | 當視訊串流在有資料遺失的環境中傳輸時,會影響到接收端的解碼品質,常見的解法是利用位元率-失真最佳化(Rate-Distortion Optimization)的方式決定編碼流程中的區塊模式,以增進視訊的容錯性。但這種作法在預測接收端所產生的失真,以及對於區塊模式間搜尋最佳解的時間,皆會使編碼端的運算複雜度大幅提升,導致所產生的視訊串流無法應用於即時傳輸中。以往的研究中,大多針對上述兩個問題的特性各自設計演算法以降低運算時間,卻限制了實際上的效果。有鑑於此,本論文提出降低複雜度之演算法,以解決上述兩種問題並同時達到強健性區塊模式的選擇。首先我們重新建立端點間預測模型,並應用在以位元率-失真最佳化決定區塊模式的方法中。透過結果分析,證明此架構與傳統決策相比的確有較佳的視訊品質。同時也觀察到,被編成SKIP 模式的區塊在各種遺失的情況下皆佔有一定比例。因此本論文利用這個特性以及數種與SKIP 模式相關的決策因素來設計演算法,其主要作法為在一張畫面真正被編碼前,事先決定出可能為SKIP 模式的區塊,來減少真正需要被編碼區塊數量,以降低整體運算時間。此外,此演算法也從兩種方向減少端點間失真的估測時間。其一為將原本以像素為基準的估測轉換為以區塊為基準,其二則是在預測時間加入判斷,使得只有非SKIP模式的區塊才需做估測。為了測試可行性,我們也將所提出的演算法實作在XviD視訊編碼器中。透過實驗結果發現,降低複雜度的演算法可減少最多37% 的運算時間,且視訊品質相當於最佳區塊模式決策所產生的結果。 Video quality suffers from significant degradation when transmitted over lossy networks, where a widely accepted technique is rate-distortion optimized mode decision for error resilience coding. However, the estimation of end-to-end distortion introduces noticeable computing complexity. In addition, the search for the optimal mode among all possible modes also takes a lot of encoding time in a general video encoder. The complexity due to the two problems therefore makes it difficult to achieve robust video streaming for real-time transmission over lossy networks. Although some heuristic methods have been developed to reduce complexity, they have been designed for coping with one of the above reasons alone without consideration of each other. To address the issue, in this thesis we design a complexity-reduction algorithm which can solve the above problems to achieve robust mode decision. First, we construct a unified end-to-end distortion model and then apply it in a rate-distortion optimized framework. Through result analysis, we demonstrate that such a framework can indeed perform robust mode decision and a significant proportion of MBs are encoded as SKIP mode for all loss conditions. By using this characteristic and some factors of robust mode decision as decision criterion, we develop an algorithm which can determine the modes of possible MBs into SKIP mode before real encoding process. In addition, the algorithm reduces the time of estimation for end-to-end distortion by modifying coding structure into MB-level and calculating only for non-SKIP MBs. Finally, we implement this algorithm into XviD codec to demonstrate the practicality. Experimental results show that the proposed algorithm can reduce coding complexity up to 37% while maintaining similar average picture quality in comparison with the conventional rate-distortion framework for robust mode decision. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/40665 |
全文授權: | 有償授權 |
顯示於系所單位: | 電信工程學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-100-1.pdf 目前未授權公開取用 | 2.54 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。