透過您的圖書館登入
IP:100.24.20.141
  • 學位論文

以矩陣為基礎之離散餘弦轉換分割與合併方法及其在多媒體處理上之應用

An Efficient Matrix-Based DCT Splitter/Merger and Its Applications in Multimedia Processing

指導教授 : 吳家麟

摘要


由於終端用戶(End-user)的網路環境、電腦效能、顯示裝置及播放工具各異,因此,媒體內容提供者(Content Provider)針對同一份影片往往需要準備多種不同的版本以符合眾多用戶的需求,卻也同時造成了儲存空間的浪費。透過即時的影片格式、大小互相轉換(Video Transcoding)的技術,媒體內容提供者只需要準備一份檔案,便可以針對使用者的需求進行檔案格式或是大小等的轉換,從而達到全方位多媒體存取(Universal Multimedia Access)的目的。 近年來,因為DVD 及HDTV 的風行,以MPEG-2 壓縮的影片不斷增加,進而使得 MPEG-2 影片廣泛流通,然而,目前最新的影片壓縮標準則是由ITU-T及ISO/IEC 所共同制訂的 AVC/H.264。AVC/H.264。AVC/H.264 採用的離散餘弦轉換除了核心大小為MPEG-2 所採用的四分之一外,與 H.263+ 及 MPEG-4 Part 2 相比,在同樣的影片品質下,AVC/H.264 壓縮率可以有1.5 到2 倍的進步,這代表AVC/H.264 可以使用更少的資訊量來表示同樣的影片品質,是一個更有效的壓縮方法。因此,我們相信,無論是將影片格式由 MPEG-2 轉成 AVC/H.264 以降低檔案大小,或是將 AVC/H.264 的影片轉成最廣泛流通的 MPEG-2 影片,這兩種方向的格式轉換都是絕對必要的! 此外,目前有愈來愈多的處理器提供特別指令集來加速多媒體的應用程式,如: Intel 提出的 SIMD (Single Instruction Multiple Data)便可以有效降低矩陣計算的複雜度。因此,本篇論文提出以矩陣計算為基礎之離散餘弦轉換分割與合併方法及其在MPEG-2 與 AVC/H.264 之轉碼器與快速影像縮放之應用,經由實驗數據分析,本篇論文所提出的方法不僅更有效率,也可以維持較好的影像品質,軟硬體實作時均滿足標準規格之要求。

並列摘要


An efficient method for splitting an N×N 2-D DCT block into four N/2×N/2 or two N × N/2 (or N/2×N 2-D DCT blocks is presented in this dissertation. The merge case (the reverse direction) can be realized by inverse transposing the given matrix operations in the ”split” case. The computational complexity of the proposed methods is lower than that of the direct approach and is the same as that of the most efficient conversion approach described in the literature. Moreover, the proposed DCT Splitter/Merger can be readily implemented by specific multimedia instruction set available today. The performance of the proposed algorithm can be justified by examining the reproduced image quality. When N = 8, our method can be applied to act as a transcoder between the latest video coding standards AVC/H.264 and the older ones, such as MPEG-1, MPEG-2 and MPEG-4 part 2. Based on the DCT Splitter/Merger, the proposed algorithm also can be applied to do image resizing in the compressed domain.

參考文獻


[1] R.L. de Queiroz, ”Fast downscaled inverses for images compressed with Mchannel lapped transforms,” IEEE Trans. on Image Processing, vol.6, no.6, pp.794-807, June 1997.
[2] V. Bhaskaran, K. Konstantinides, and G. Beretta, ”Text and image sharpening of scanned images in the JPEG domain,” IEEE International Conference on Image Processing, Santa Barbara, CA, Oct. 1997.
[3] S.F. Chang and D.G. Messerschmitt, ”Manipulation and compositing of MCDCT compressed video,” IEEE J. on Select. Areas Commun., vol.13, no.1, pp.1-11, Jan. 1995.
[5] K.R. Rao and P. Yip, ”Discrete Cosine Transform: Algorithms, Advantages and Applications,” Academic Press, 1990.
[9] H. Kalva, ”Issues in H.264/MPEG-2 Video ranscoding,” IEEE International Conference on Consumer Communications and Networking, Las Vegas, NV, Jan. 2004.

延伸閱讀