- 學位論文
自適性串流架構下之視訊影像品質最佳化設計與實作
Design and Implementation of Quality Optimization Strategy in Adaptive Streaming
摘要
受到智慧型終端的崛起與網際網路的廣泛運用，視訊串流成為一個近期備受關注之熱門技術。MPEG-DASH允許標準化之客戶端與伺服器端傳遞視訊串流服務，其服務能依照網路狀態的改變動態調整視訊串流之畫面品質，進而提供流暢且高品質之視訊串流體驗。 本論文實作一套完整之MPEG-DASH系統，並且使用開源式軟體建置包含伺服器、網路通道以及客戶端的系統環境。我們同時提出一套新的MPEG-DASH的自適性品質決策機制，此自適性機制能夠依照當下之網路狀態與暫存器使用量選擇最適當之視訊片段，以提供順暢且穩定之視訊串流服務。而本論文方法也在與VLC的品質決策機制模擬比較中，不論是客觀性與主觀性的分析上都有更好的效果。
並列摘要
Video streaming has become an enabling technology for multimedia applications, owing to the popularity of mobile devices and wide adoption of the Internet. MPEG Dynamic Adaptive Streaming over HTTP (DASH) allows a standard-based client to stream multimedia content from any standard-based server. The adaptation mechanism in DASH can switch quality levels according to network conditions. In this paper, we implement a complete MPEG-DASH system, which includes the server, channel, and client, exclusively by open-source software. We also propose a new decision mechanism for quality control with MPEG-DASH. The adaptation algorithm selects the most appropriate video segment according to network bandwidth and buffer usage. The emulation results substantiate the superiority of the proposed method over the quality-control method given in VLC, in both visual quality and network efficiency.