H.264之多幅參考畫面演算法則之研究; Fast Multi-frame Motion Estimation Algorithm in H.264/AVC

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Communication Engineering > Electronic Thesis & Dissertation > Item 987654321/8176

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/8176

Title:	H.264之多幅參考畫面演算法則之研究;Fast Multi-frame Motion Estimation Algorithm in H.264/AVC
Authors:	王家任;Jia-Ren Wang
Contributors:	通訊工程研究所
Keywords:	多幅參考畫面;移動估計;模式決策;搜尋視窗;預測向量;H.264;可變區塊大小;search window;multi-frame;motion estimation;predictive motion vector;H.264
Date:	2007-07-06
Issue Date:	2009-09-22 11:20:11 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	現今對於多媒體影音方面的要求越來越大，且畫面品質要求也越來越高，因此多媒體影音格式的制定一直是很熱門的話題和研究。逐漸崛起的影音格式H.264 利用很多不同以往的編碼和預測方式，可容易達到高品質且壓縮位元率的目的。但也因此需要付出很重的運算複雜度來達成，因此減少複雜度且可維持一定品質的方法，即為我們此論文的探討內容。我們首先提出不同以往架構的多幅參考畫面候選模式的演算法，可使其計算量可以節省30%左右，然後又運用比較精準的向量預測方式，加上可變搜尋視窗的方法，來達到縮小搜尋視窗的目的，又可再多節省不少時間的運算，平均差不多50%左右的時間節省，最後結合所有的方法，可達到節省70%左右。實驗結果顯示我們不僅可達到一定時間的節省，但是畫面品質也可在可接受範圍的損失，所以運用到即時(Real Time)系統上運用，應該是可行的方法。 The emerging popular video coding standard, called H.264/AVC, supports many advanced techniques to achieve better performance compared to the previous one. As the performance got better and better, the computation load must be paid as return. In order to make it apply in the real-time system as possible, so many researched results had been proposed to reduce the complexity. In this thesis, we also proposed some algorithms to achieve the same goal. First, we proposed how to distinguish candidate modes from the others and terminated other frames of the candidate modes. And then we combined the predictive motion vector with the adaptive search window to shrink the search process. Finally, we integrated these three proposed methods into the whole novel multi-frame motion estimation algorithm. The simulation result showed that we could save for 70% in coding time with insignificant performance loss.
Appears in Collections:	[Graduate Institute of Communication Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...