適用於易出錯網路通道之多重描述視訊編碼

Abstract

無論在網際網路或無線網路中，封包遺失以及突發性錯誤都會對視訊傳輸系統，例如視訊會議，點對點串流傳輸等系統，造成視訊品質的傷害，多重描述視訊編碼即是被提出解決網路不穩定所導致問題的方法。 在此篇論文中，提出兩種產生描述的方法，其中一個修改自多重描述量化編碼，主要針對減少多餘編碼量和增加編碼彈性設計，另外一個方法則是針對頻率域係數分配方法的改善，使描述間可達到平衡。而提出的估測機制能有效的減輕因為描述遺失而造成的視訊解碼錯誤傳播的現象。實驗結果證明此篇論文所提出的方法能有效的增進多重描述編碼的效率。

Packet loss and burst errors occur in the internet or wireless network might cause serious consequence of quality degradation for video applications such as peer-to-peer video streaming, video conference, etc. Multiple description video coding has been proposed to combat the problems caused by error-prone network channels. In this paper, two methods of description generation have been proposed. One method is a modified multiple description scalar quantizer which is intent on redundancy reduction and the coding scalability. The other is the coefficient partition method in transform domain, in such a way that the descriptions could be more balanced. And an estimating mechanism is proposed to alleviate the drifting problem caused by the description loss by synchronizing the reference frames at the encoder and the decoder. The experiments show that the proposed method has a significant improvement at the event of description loss.