標題: 使用R-D最佳化之多重描述量化編碼
Rate Distortion Optimized Multiple Description Scalar Quantization
作者: 楊巧安
Yang, Chiao-An
蔡文錦
Tsai, Wen-Jiin
多媒體工程研究所
關鍵字: 多重描述編碼;R-D最佳化;多重描述量化編碼;Multiple description coding;rate-distortion optimized;multiple description scalar quantization
公開日期: 2012
摘要: 隨著資訊的發達以及網路的發展,數位資訊經由網路傳輸的需求持續的增加,當視訊影片在不穩定的網路傳輸時,不可避免會出現封包遺失的情況。因此,如何減少封包遺失對視訊影片得影響,是非常重要的一個研究議題。 多重描述編碼(Multiple Description Coding)是一種編碼技術用來降低因為傳輸錯誤所造成的品質傷害。在本篇論文中,我們提出了一個應用在多重描述量化編碼的R-D優化編碼方式,利用各個區塊的內容以及封包遺失率來決定各個區塊要使用的編碼策略,並將各區塊的編碼模式傳送至解碼端。解碼端可利用這些編碼模式來決定解碼方式或是在有封包遺失時進行簡單的重建。透過實驗證實,我們提出的方法在封包遺失較少的時候能夠獲得較好的編碼效能。
Multiple description video coding (MDC) [1] is one of popular solutions to reduce error propagation caused by transmission over error-prone networks. In this paper, a rate-distortion optimized coding based on multiple description scalar quantization (MDSQ) is proposed. With the proposed method, video contents and channel conditions are used to determine the rate-distortion optimized (RDO) mode, and adjust the encoding strategy. At decoder side, the decoder determines the decoding strategy according to RDO modes and packet loss conditions. Experimental results showed that the proposed method effectively improves the coding performance for low packet loss rates such as 1%~ 5%.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079957516
http://hdl.handle.net/11536/71927
Appears in Collections:Thesis