欢迎访问《图学学报》 分享到:

图学学报

• 视觉与图像 • 上一篇    下一篇

基于DS-CDS 的改进运动估计算法及实验

  

  • 出版日期:2015-06-24 发布日期:2015-06-29

Improvement of Motion Estimation Algorithm and Experiment Based on DS-CDS

  • Online:2015-06-24 Published:2015-06-29

摘要: 为了提高视频的压缩效率,在菱形搜索算法和十字菱形搜索算法的基础上,结合实
际运动图像中的运动向量以水平方向向量为主的特点,提出了一种利用偏水平十字模板搜索与偏
向双菱形模板搜索相结合的改进搜索算法。为检验本文改进算法的效果进行了对比实验,结果表
明:本文提出的基于偏水平十字及偏向双菱形搜索法适合各种运动类型的视频序列,更适用于运
动变化剧烈的序列,并且能够在PSNR 值和BR 值接近最优水平时,大大减少运动估计时间,相
比于FS 算法,对QCIF 格式图像的运动估计时间减少约95%,对CIF 格式图像的运动估计时间
减小约94%。

关键词: 视频压缩, H.264, 运动估计, 菱形搜索算法, 十字菱形搜索算法

Abstract: To improve the compression efficiency of the video, based on diamond search algorithm and
cross diamond search algorithm, combined with the feature that motion vector was mainly focused on
horizontal direction vector in actual motion images, the thesis put forward an improved search
algorithm which combined partial to horizontal cross template search with biased toward double
diamond template search. Comparative experiments were conducted to test the effect of improved
algorithm. The result of the performance contrast experiment shows that the thesiss improved
algorithm suits all kinds of motional video sequences, especially those sequences changing poignantly
in movement. Under the condition that the PSNR and the BR value are very close to the optimal level,
compared to the FS algorithm, the improved algorithm decreases approximately 95% of the motion
estimation time of the QCIF pictures, and decreases about 94% of the motion estimation time of the CIF
pictures, thus decreases greatly the motion estimation time.

Key words: video compression, H.264, motion estimation, diamond search algorithm, cross diamond
search algorithm