透過您的圖書館登入
IP:3.137.175.113
  • 期刊

用於H.264的快速inter與intra模式決策演算法

Fast Inter/Intra Mode Decision In H.264

摘要


在過去移動預估,使用固定區塊大小16*16做預估,因為影片序列特性上的不同,會造成移動向量預估不夠精確。H.264/AVC提出可變區塊大小模式來解決此問題。此模式分成七種不同區塊大小,可以因應不同的影片序列特性,根據不同特性選擇不同的模式做預估,因此精確度會提高。但因為每種模式都要做判斷,最後由程式決定出一個最佳的模式,所以複雜度相對提高。為了改善複雜度問題,過去針對移動預估提出以下幾種方式,包含邊緣計算之Sobel方式、DCT計算方式、移動向量分割與合併的演算法、2*2濾波器計算不同角度之快速intra模式。因此如何在維持其品質的情況下,儘量減少運算量,加快編碼速度,以達到及時傳輸目的。為了要徹底改善複雜度問題,本論文提出以移動向量為基準之快速intra/inter模式決策演算法。此方法不用額外計算,基本上是利用移動向量統計出來的數值來做快速模式預估,因此可以省略很多不必要模式選擇。由於影像序列是由不同移動向量所組成,且移動向量與不同特性影片序列相關性非常高,所以影片序列可以根據影片特性切割成快、慢和靜止區塊。在實驗中,我們以七種不同影像序列做移動向量與影片序列特性之間的分析。實驗環境是在JM12.2版本,每個序列採用前100張畫面共七種影像序列。就實驗結果而言,使用inter模式預估,時間上約可節省70%,畫面品質約下降0.04db,位元率約上升0.7 kbit/s,都比相關論文所提方法來的快,畫面品質和位元率也都有一定維持。而使用intra 模式預估時間上約可節省74%,品質約下降0.02db,位元率約上升3.0 kbit/s。

並列摘要


In past motion vector estimation, constant 16*16 block size is usually used for prediction. Due to different characteristics of video sequences, fixed block size is not adequate to predict motion vector accurately. H.264/AVC proposes the mode of variable block size to increase accuracy of motion vector prediction according to the characteristics of video sequences. However, complexity is relatively high because each mode must be judged. In order to decrease complexity and greatly enhance compression efficiency of H.264/AVC, a fast inter/intra mode decision algorithm in H.264 by motion vector prediction is proposed in this paper. We make analysis toward motion vector and sequence characteristics in seven sequences. Under experimental environment of JM12.2 version, totally 700 frames are used in seven sequences. Experimental result shows that inter prediction could save 70% of time with image quality decreasing about 0.04db and bit rate increasing about 0.7 kbit/s, all of which are superior than methods proposed in other relevant paper with stable image quality and bit rate. However, even though intra prediction saves 74% of time, quality still decreases 0.02db and bit rate increases about 3.0 kbit/s. How to correct this higher bit rate should be considered in improvement issue in future.

延伸閱讀