首页> 中文期刊>西安电子科技大学学报(自然科学版) >用于非规则形状宏块划分的位移矢量预测方法

用于非规则形状宏块划分的位移矢量预测方法

     

摘要

To improve the accuracy of predicting the motion vector for irregularly-shaped motion compensation partition, a novel motion vector prediction method based on the adjacent length of neighbor motion compensation partition (MVPAL) is proposed with the consideration of the spatial correlation between the motion information. Firstly, according to the conclusion that the correlation between the motion information strengthens the adjacent length, the proposed method predicts the motion vector with the motion information on the neighboring motion compensation partition with the largest adjacent length. Then, the current coding motion information is stored in the 4x4 block manner for predicting the motion vectors of motion compensation partitions in the remaining macroblocks. Experimental results demonstrate that MVPAL outperforms the existing algorithm for motion vector prediction with different resolution sequences. The average coding efficiency has been improved up to 1.47% compared with the original motion vector prediction algorithm in H. 264/AVC.%为了提高非规则形状运动补偿单元的位移矢量预测的准确性,去除运动信息的空间冗余,提出一种基于邻边长度的位移矢量预测方法.首先,根据邻边长度越长,相邻运动信息间的相关性越强的分析结论,选择最大邻边长度运动补偿单元的运动信息,用于当前非规则形状运动补偿单元的位移矢量预测.然后,将最佳位移矢量按照4×4块规则位置统一存放,以便预测后续编码宏块中运动补偿单元的位移矢量.实验数据表明,与H.264/AVC中位移矢量预测方法相比,本算法编码效率平均提高1.47%.

著录项

相似文献

  • 中文文献
  • 外文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号