針對多視角視訊編碼固定圖框之全域位移向量估測與區塊模式決策加速機制

Fast Global Disparity Vector Estimation and Block Mode Decision for Anchor Frame in Multiview Video Coding

葉錦坤(Chin-Kun Yeh)

指導教授 ： 郭天穎

國立臺北科技大學/電資學院/電機工程系研究所/碩士(2008年)

https://doi.org/10.6841/NTUT.2008.00327

摘要

多視角視訊壓縮(Multiview Video Coding, MVC)於2006年7月開始被列入JVT (Joint Video Team) H.264延伸計畫中，之後於2007年10月多視角視訊亦被加入ITU網路電視(IPTV)服務選項，由此可見多視角系統在未來的重要性。由於H.264 MVC系統不僅採用H.264編碼且包含了視角上的參考，以提高了壓縮效率，卻也大幅地提升編碼複雜度，因此有效地降低MVC計算複雜度為重要的研究議題。然而先前文獻上有關多視角視訊編碼的運算量化簡演算法，多透過視角間之關係化簡時軸上運動估測時所需運算量，較少針對視角間參考編碼時之複雜度作化簡。因此本研究提出在運動跳躍模式(Motion Skip Mode, MS)開啟之下，編碼固定圖框(Anchor picture)時，利用邊緣偵測之投影量估算其全域位移向量(Global Disparity Vector, GDV)，並利用GDV資訊有效率地執行固定圖框區塊模式決策，以加速MVC視角間之編碼流程。實驗結果證明本研究提出之方法於固定圖框編碼中可節省運算複雜度且維持良好R-D效能。

關鍵字

多視角視訊壓縮 ； 運動向量跳躍模式 ； 固定圖框 ； 全域位移向量 ； 區塊模式決策

並列摘要

In July 2006 the Joint Video Team (JVT) started working on the multiview video coding (MVC) extension for H.264 video. In Oct. 2007 multiview video has also been added as an option in ITU Internet protocol TV (IPTV) service. So it shows that multiview video system is getting more attention in the near future. The H.264 MVC extension achieves excellent coding efficiency by not only inheriting from H.264 video standard but also incorporating new tools for inter-view prediction. As a result, a huge amount of computational complexity is involved, and how to reduce the complexity of encoding process becomes an important issue. In previous work, most researchers proposed to expedite the motion estimation across the temporal axis by utilizing the inter-view relationship to save the computational load on MVC, but few of them considered to reduce the complexity spent in the inter-view coding. In this thesis, we speed up the MVC inter-view coding by proposing a fast global disparity vector (GDV) estimation based on the edge projection map, and also present a fast block mode decision for the MVC anchor frames based on the obtained GDV with motion skip (MS) mode enabled. Experimental results show that our proposed method can reduce the complexity while keeping R-D performances for MVC anchor frame encoding.

並列關鍵字

Multiview video coding (MVC) ； Motion skip mode ； Anchor frame ； Global disparity vector (GDV) ； Mode decision

參考文獻

[1] M. Tanimoto “Overview of Free Viewpoint Television,” Proc. IEEE Conf.on Image Processing, vol. 21, no. 6, July 2006, pp. 454-461.

[2] I. Ahmad, “Multi-View Video: Get ready for Next-Generation Television,” IEEE Distributed System, vol. 8, Issue 3, March 2007, pp. 1-6

[4] T. Fujii and M. Tanimoto, “Free-Viewpoint TV System Based on Ray-Space Representation,” SPIE ITCom , vol. 4864, Aug. 2002, pp. 175-189.

[5] “Multi-View Video,” Available:

[6] S. B. Kang, R. Szeliski, and P. Anandan, “The Geometry-Image Representation Tradeoff for Rendering,” Proc. IEEE Int. Conf. on Image Processing, vol. 2, 10-13 Sept. 2000, pp. 13-16.