透過您的圖書館登入
IP:18.218.101.36

摘要


本論文針對立體視訊編碼需求提出了一個編碼器設計,可支援H.264/AVC 2D視訊與MVC 3D視訊的編碼。因為利用MVC進行立體視訊編碼時需要同時處理不只一個視角,加上非基礎視角的其它每個視角還需要額外再多做視角間預測,因此計算量相較於利用H.264/AVC的傳統2D視訊而言變得十分龐大,為了克服上述問題,我們在畫面內預測、畫面間預測、視角間預測、情境適應性二元算術編碼、與去區塊效應濾波模組中提出了新的快速演算法或架構設計,可在大幅降低計算量的情況下維持與原來近似的品質。

並列摘要


This article describes an encoder design for 3D stereoscopic video coding. This encoder can support 2D video coding by H.264/AVC and 3D stereoscopic video coding by MVC. The 3D stereoscopic video coding is required to process multiple views captured at the same time instant. In addition, the inter-view prediction is performed to further reduce redundancy. Consequently, the requirement of its computing power is much greater than traditional 2D video coding by H.264/AVC. To reduce the computation time of the MVC encoder under negligible video quality decay, several novel fast algorithms and architectures are proposed for intra prediction, inter-picture prediction, inter-view prediction, CABAC, and de-blocking filter.

延伸閱讀