軟硬體共同開發即時嵌入式多媒體MPEG-4系統設計與實作

Abstract

多媒體系統是一門綜合科學包含了影像輸入、影像壓縮、數位影像處理、影像輸出等學科，按期工業界的應用可概分為數位相機及數位攝影機兩大類，數位相機的目的在保存靜態影像，數位攝影機的目的在保存動態影片，此兩者的差別在於數位攝影機多了影像壓縮學科。然而傳統的影像壓縮技術已無法滿足處理龐大影像資料的需求，尤其在高畫素影像輸入時代來臨，動輒數百萬像素，如何快速及更有效率的處理動態影像資料壓縮，成為一個刻不容緩的議題。

本文主題係改進傳統影像壓縮技術(MPEG-4)，針對演算法的部分一一探討，提出合理的質疑並藉由相關實驗來驗證我們的推論，並針對問題來源提出改善方案。將部分演算法功能硬體化，以提高執行速率之外，並建構一個多媒體系統平台，以軟體控制影像壓縮的流程以達到系統上的最佳化來增進系統效能，實驗的系統平台包括即時錄影一段影像並在顯示器上播放，筆者從線路圖設計、系統規劃、多層板佈線、嵌入式作業系統、視窗系統的程式設計等,規劃一連串的驗證流程。最後也討論未來如何改善此多媒體系統的後續研究。

Multimedia system is a comprehensive science which mergers the disciplines of image sensor, video compression, digital signal process, and image display etc. It can be categorized as digital camera and digital video according to industrial applications. The purpose of digital camera is to conserve still pictures while digital video is to record dynamic video movie. The difference between these two commercial products is that digital video requires extra video compression and depression techniques. However, traditional video compression and depression techniques can not satisfy the needs of huge image data. As the coming of high image resolution era, the subject to resolve this problem is critical.

Our thesis is to improve video compression and decompression technique (MPEG-4) proposed by John Dunlop. We discuss MPEG-4 algorithms and bring up reasonable queries and test our inference by related experiments which help us to find two better solutions. The first solution is to run some of the video compression algorithm by hardware circuit instead of software simulation to increase execution speed. The second solution is to build up a multimedia system from scratch to testify our assumption. We start this experiment from schematic design, system planning, PCB layout, embedded operating system programming, windows programming etc. We also discuss the possible future work to enhance the proposed multimedia system.