目前有許多的影像資料都是採用MPEG-2標準去儲存和傳送，像是DVD和DTV等不同方面的應用，而H.264是目前最新的視訊壓縮標準之一，它適合較低頻寬的網路傳送，與MPEG-2相比，也能達到較佳的編碼效率，既然過去的檔案資料是不能捨棄的，從MPEG-2到H.264的轉換編碼系統就變的十分重要，我們試圖去發展一個轉換編碼系統，它可以降低計算複雜度又不會造成影像品質的大量衰退。 轉換編碼系統最主要的問題，在於如何在編碼端將解碼端獲取的資訊做再利用並避免多餘的編碼程序計算，這篇論文裡面所討論的是畫面間轉換編碼，而畫面間轉換編碼包含了畫面內和畫面間的模式預測，在畫面間模式預測中，我們先擷取解碼端的資訊，將MPEG-2的影像區塊作分類，再利用零區塊偵測和快速移動估計的演算法去加速整個轉換編碼系統；在畫面內模式預測則是利用零區塊偵測和絕對誤差轉換和演算法去做加速，最後我們將這兩部分的演算法合併，提出一個整合型的快速畫面間轉換編碼演算法，實驗的模擬數據顯示，我們的演算法能大幅的縮短編碼時間，而且該演算法在影像品質上所造成的衰減也是可以忽略的。 There are many videos sources recorded and transmitted in the MPEG-2 format for DVD and DTV applications. H.264 is one of the latest video compression standards, and suitable for low-bandwidth networks such as wireless network. It can achieve better coding efficiency than MPEG-2. Since we can’t abandon the files recorded in MPEG-2 format, transcoding system from MPEG-2 to H.264 has become very important. We should develop a transcoding system which can reduce the computation complexity without reducing video quality extensively. The key problem to speed the transcoding system is how to reuse the decoded information into encoding process to avoid redundant encode computation. We will discuss the inter frame transcoding here. Inter frame transcoding includes intra mode prediction and inter mode prediction. In inter mode prediction, we retrieve decoded information in MPEG-2 and classify them into different type. Then we use ZBD detection and FME algorithm to speed up the transcoding system. In intra frame prediction, we use ZBD and SATD-based to speed up the transcoding system. In the end, we combine the two system into a conformable transcoding system. According to the simulation results, the proposed algorithm can reduced the computation time extensively. The degradation in video quality caused by proposed algorithm is negligible.
