MPEG2 動態補償元件之設計

Abstract

MPEG標準的建立開始於1988年，其主要目的是針對於視頻及相關音頻信號 的應用訂定標準。建立標準的目的是為了降低實現硬體成本及解決相容性 的問題。現行的標準有兩個版本---MPEG1，已經有多方面應用的商業產品 ；MPEG2，已有數家公司（IBM，東芝，SGS-THOMSON） 發展出晶片。雖然 其應用仍受制於其周邊硬體，但由於MPEG2標準所含括的應用範圍更廣， 更高的傳輸率及更高的解析度和畫質，將其以硬體實現已成為目前熱門研 究主題。在本論文中,我們提出一個可以應用在 MPEG2 中動態補償元件 ( Motion Compensation Module ) 的架構,此架構可以用來還原 MPEG2中 的影像,此影像的移動向量 ( Motion Vector ) 的精確度甚至可以達到二 分之一點 ( pixel ),而且此架構可以處理 MPEG2 中任何的預測模式 ( prediction mode )。另外我們也說明了在計算機輔助設計環境下,電路設 計技巧和設計流程。在設計流程中，我們簡單地介紹設計方法以及如何驗 證和最佳化所設計的電路。在論文最後,我們便利用上述所提的架構和設 計流程,用超大型積體電路 ( VLSI )的方法,將此動態補償元件設計出來 MPEG standard has been established since 1988 for coding of moving pictures and associated audio applications. The goal of establishing this standard is to reduce the implement cost and solve the problem of compatibility. There are two current versions of MPEG standard -- MPEG1, commercial products have been used in many applications; MPEG2, several companies(IBM, Toshiba , SGS THOMSON) have produced MPEG2 chips. although the applications are still limited by other hardwares, the hardware implementation of MPEG2 is a very popular topic for industry due to its wide range of applications, high bit rates, good resolutions and quality. In this thesis, we present an architecture for the MC ( Motion Compensation )module of the MPEG2 decoder. It can predict a frame whose motion vectors are specified to an accuracy of one half sample. This architecture can reconstruct a frame ( field ) predicted in any prediction modes. We also present the circuit techniques and design flow under Synopsys environments. In the design flow, we briefly discuss the design methodology and verification approach of optimization of verilog behavior description circuits. Finally, we implement this MC/Interpolator module based on the architecture and the design flow.