Zhang Qiuwen, Li Nana, Wu Qinggang
College of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou 450002, China.
ScientificWorldJournal. 2014;2014:620142. doi: 10.1155/2014/620142. Epub 2014 May 19.
The emerging international standard of high efficiency video coding based 3D video coding (3D-HEVC) is a successor to multiview video coding (MVC). In 3D-HEVC depth intracoding, depth modeling mode (DMM) and high efficiency video coding (HEVC) intraprediction mode are both employed to select the best coding mode for each coding unit (CU). This technique achieves the highest possible coding efficiency, but it results in extremely large encoding time which obstructs the 3D-HEVC from practical application. In this paper, a fast mode decision algorithm based on the correlation between texture video and depth map is proposed to reduce 3D-HEVC depth intracoding computational complexity. Since the texture video and its associated depth map represent the same scene, there is a high correlation among the prediction mode from texture video and depth map. Therefore, we can skip some specific depth intraprediction modes rarely used in related texture CU. Experimental results show that the proposed algorithm can significantly reduce computational complexity of 3D-HEVC depth intracoding while maintaining coding efficiency.
新兴的基于高效视频编码的3D视频编码(3D-HEVC)国际标准是多视图视频编码(MVC)的后续标准。在3D-HEVC深度帧内编码中,深度建模模式(DMM)和高效视频编码(HEVC)帧内预测模式都用于为每个编码单元(CU)选择最佳编码模式。该技术实现了尽可能高的编码效率,但导致编码时间极长,阻碍了3D-HEVC的实际应用。本文提出了一种基于纹理视频与深度图相关性的快速模式决策算法,以降低3D-HEVC深度帧内编码的计算复杂度。由于纹理视频及其相关深度图表示相同场景,纹理视频和深度图的预测模式之间存在高度相关性。因此,我们可以跳过相关纹理CU中很少使用的一些特定深度帧内预测模式。实验结果表明,该算法在保持编码效率的同时,能够显著降低3D-HEVC深度帧内编码的计算复杂度。