應用於三維高效率視訊編碼之基於深度資訊與時空相關性的快速模式決定演算法

A Fast Mode Decision Algorithm Based on Depth Map and Spatial-Temporal Correlation for 3D HEVC

李明璋(Ming-Chang Li)

指導教授 ： 吳家麟

國立臺灣大學/電機資訊學院/資訊網路與多媒體研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.02592

摘要

多視角影片(Multi-view Video)是近年新興的影片格式，相較於傳統影片格式，多視角影片通常伴隨著相對應的深度資訊(Depth Map)，用來表達場景中物體與攝影機的距離關係。因此進行影片編碼時，可以利用這些深度資訊進一步提升壓縮效能，但同時也會增加許多計算複雜度。

高效率視訊編碼(High Efficiency Video Coding, HEVC)是目前最新的視訊編碼標準，由ISO/IEC MPEG與ITU-T VCEG組成的JCT-VC所制定。三維高效率視訊編碼(3D HEVC)是其延伸，並針對多視角影片的特性所提出的新壓縮標準。由於HEVC和3D HEVC都是以壓縮高解析度影片為目的，在壓縮模式決定過程(Mode Decision Process)中，會比先前的標準嘗試更大且更多種類的壓縮單元(Coding Unit)與壓縮模式(Coding Mode)，因此整體耗費的時間遠超過先前的視訊壓縮標準。

為了使3D HEVC能更加普遍地適用於各種應用，本篇論文嘗試利用多視角影片所特別具有的深度資訊，輔以空間和時間上的相關性，縮短壓縮模式決定過程的執行時間，進而提升3D HEVC壓縮時的速度。實驗結果顯示，本篇論文提出的快速模式決定演算法(Fast Mode Decision Algorithm)平均可以縮短3D HEVC 60%以上的壓縮時間，同時不會使壓縮效能受到明顯的影響。

關鍵字

多視角影片 ； 深度資訊 ； 高效率視訊編碼 ； 三維高效率視訊編碼 ； 快速模式決定演算法

並列摘要

In recent years, multi-view videos attracted lots of attention in video coding communities. Compare to the traditional 2D videos, multi-view videos usually contain an additional information, the depth map, which stores the distance between scenes and the camera position. When coding multi-view videos, the depth map information can be utilized to further increase the coding efficiency as well as reducing the computational complexity.

High Efficiency Video Coding (HEVC) is the latest standard of video coding organized by ITU-T Video Coding Experts Group (VCEG) and ISO/IEC Motion Picture Experts Group (MPEG). Three-Dimension High Efficiency Video Coding (3D HEVC), which is customized for the compression of multi-view videos, is one of the extensions of HEVC. Since both HEVC and 3D HEVC are designed to compress high-resolution videos, the coding units and coding modes are much larger and more complex than previous standards. Therefore, the overall executing time increases significantly than that of the previous standards.

In order to apply 3D HEVC to practical applications, its inherent high complexity must be largely reduced. In this thesis, a fast mode decision algorithm for 3D HEVC is proposed based on spatial and temporal correlations and depth map information. According to the experimental result, the proposed algorithm can reduce about 60% executing time of the overall encoding process, and the loss in coding efficiency is negligible.

並列關鍵字

Multi-view Video ； Depth Map ； High Efficiency Video Coding (HEVC) ； Three-Dimension High Efficiency Video Coding (3D HEVC) ； Fast Mode Decision Algorithm

參考文獻

[4] A. Vetro, T. Wiegand, and G. J. Sullivan, “Overview of the Stereo and Multiview Video Coding Extensions of the H.264/MPEG-4 AVC Standard”, Proc. IEEE, Vol. 99, No. 4, pp. 626-642, Apr. 2011.

[6] J. Leng, L. Sun, T. Ikenaga, S. Sakaida, "Content Based Hierarchical Fast Coding Unit Decision Algorithm for HEVC", IEEE Int Conf. Multimedia and Signal Proc., pp.56-59, May 2011.

[7] L. Shen, Z. Liu, X. P. Zhang, W.Q. Zhao, and Z.Y. Zhang, “An Effective CU Size Decision Method for HEVC Encoders”, IEEE Trans. Multimedia, Vol. 15, No. 2, pp. 465-470, Feb. 2013.

[9] L. Shen, Z. Liu, T. Yan, Z.Y. Zhang and P. An, “Early SKIP Mode Decision for MVC Using Inter-view Correlation”, Signal Process: Image Commun., Vol. 25, No. 2, pp. 88-93, Feb. 2010.

[10] W. Wei, P. Chen, Y. Zheng, and J. Feng, “Fast Inter-view Prediction and Mode Selection for Multiview Video Coding”, 4th International Congress on Image and Signal Processing, pp. 151–156, Oct. 2011.