分布式视频编码中一种基于混合模型的边信息生成算法
Hybrid Model-Based Side Information Generation Algorithm for Distributed Video Coding
-
摘要: 提出了一种基于块匹配模型和光流模型的混合式边信息生成方法.该方法对块匹配效果较差的区域采用光流法重新估计其运动矢量,得到了更为精确的运动场.实验表明,该算法生成的边信息质量明显高于传统的基于块匹配的运动估计内插法.Abstract: A hybrid block-based and optical flow-based method is proposed.By using the optical flow method to recompute the motion vector in the regions where block matching algorithms have poor performance,this method produces more accurate motion fields.The experimental results show that comparing to other algorithms,the proposed method can significantly improve the quality of side information.
下载:
