摘要
提出一种基于深度的虚拟视失真估计方法,将深度图宏块划分为平坦宏块和非平坦宏块,结合各自特点分别用频域块级方法和时域像素级方法估计由深度失真引起的虚拟视失真。实验证明,该方法估计准确度较高,且复杂度比像素级模型有所降低。它对于提高深度视频编码效率和优化多视+深度(MVD)码率分配具有重要指导意义。
A depth based virtual view distortion estimation method is proposed in this paper. The coding macro-blocks in depth map are classified into even macro-blocks and uneven macro-blocks. The block-level frequency domain distortion estimation method is used in even macro-blocks while pixel-level time domain method in uneven macro-blocks. The advantages of the two estimation methods are combined effectively. The simulation experiments confirme that the proposed method can estimate the virtual view distortion caused by depth map distortion with good balance between calculate accurate and compute complexity. It provides importance to improve 3DV code efficiency and rate allocation performance.
出处
《电视技术》
北大核心
2014年第11期215-218,223,共5页
Video Engineering
基金
国家自然科学基金重点项目(60832003)
国家自然科学基金项目(61172096)
上海市科委项目(12DZ2293500)
