新一代的H.264 不管是壓縮率和影像品質都明顯優於以往的視訊標準｡這是因為其使用了新的技術可變區塊大小的移動估計、多幅參考畫面的移動估計、整數轉換｡ H.264運用了多幅參考畫面移動估計的技術,這技術讓H.264 編碼效益提升許多,但同時也增加了龐大的運算複雜度,我們提出一種演算法基於連續畫面間的相關性以改進多幅參考畫面移動估計之編碼效率｡ 在本章中,我們修改了幾個先前的演算法,並提出了一種可調適機制,可以利用移動特性來改善先前的作法｡在提出的演算法中,我們使用一個簡單的標準分類所有的巨區塊特徵 (穩定區域或非穩定區 域)｡ 實驗結果顯示,建議的演算法對多幅參考畫面全搜索演算法的比較可以顯著減少計算複雜度,同時保持了良好的編碼性能｡ The new video coding standard, H.264/AVC, uses variable block size ranging from 4x4 to 16x16 and multi-reference frame in interframe coding｡ We proposed an effective algorithm to improve the multiple reference frame motion estimation scheme in H.264,based upon correlation among successive frames and neighboring blocks｡ To improve coding efficiency, in this paper, we modify the previous algorithm and propose an adaptive mechanism, which makes use of motion characteristic to modify previous work｡In proposed algorithm, we make use of a simple criterion to distinguish the macroblock characteristic(stationary region or non-stationary region)｡ The experimental results reveal that the proposed algorithm brings out a significant reduction in computation compared to the multiple reference frame full search algorithm, while maintaining a good coding performance｡
