多媒体资讯为能在拥挤的共享网跟稳定的无线网上传输资料.pdf

摘要 多媒體資訊為了能在擁擠的共享網路和不穩定的無線網路上傳輸，常使用 多重描述編碼法(Multiple Description Coding, MDC) ，將多媒體資料分解成可獨立 解碼的碼流，並透過傳輸通道分別傳送，以增加多媒體資訊編碼傳輸的可靠性與 穩定性。本論文提出改進 MDC編碼架構的方法: (1) 將視訊分成兩個獨立的子視 訊，以 MC-EZBC的編碼技術來產生兩個品質可調之碼流 ; (2)利用兩個子視訊間 的相關性，運用 DVC(Distributed Video Coding)技術將冗餘信號有效的編碼; (3) 在子訊號碼流與 DVC 之間，利用預測式編碼架構，充分利用影像之間的關聯性， 在相同的位元率下提高 PSNR的效能 ;(4)將一個子訊號碼流與一個DVC碼流結合 成一個描述碼流(Description)並以FEC(Forward Error Correction)對Description 進 行錯誤回復的保護機制來分別傳送兩的描述碼流。本論文我們將 MDC 與 DVC 兩種技術有效的整合，並且提出改進編碼傳輸的方法，使得接收端的有高品質與 高穩定性的視訊。實驗結果顯示，相較於現有的方法，運用本法可以提高訊雜比 約0.5~1 dB 。 i Abstract Multiple Description Coding (MDC) is an effective solution of multi-media transmission in the jammed internet and unstable wireless networks. MDC promises the stability and reliability of multi-media communication with multiple transmission paths. We have proposed an improved MDC architecture that: (1) decomposes the original video into two sub-videos which are encoded using MC-EZBC coder; and (2) correlations between sub-videos are exploited and encoded using the Distributed Video Coder (DVC); (3) Apply predictive coding framework between sub-video and DVC bit stream; and well exploit the correlations between video to enhance PSNR at the same bit rate;(4) combine the encoded data of above two parts to generate one description. The generated two descriptions are transmitted independently and are protected by FEC (Forward Error Correction) algorithm. The PSNR of reconstructed video can be improved to 0.5 to 1 dB higher, as compared to previous researches. Experiments also show that, by effectively integrate residual DVC coding with MDC, the stability and quality are la