摘要
目的通过分析经典运动估计算法,并归纳其技术要点比较总结其优缺点,提出一种快速运动估计算法。方法利用小菱形(SDSP)模板和六边形(HSP和VSP)组合模板自适应地进行模板匹配搜索。结果充分利用序列图像的时间、空间相关性进行运动类型的划分,并进行了起点预测,实现了快速运动算法。结论仿真实验证明该改进算法在保证重构图像质量基本不变的情况下,减少了运动搜索点数,提高了运动搜索速度,缩短了运动搜索的时间。
Aim By analyzing classical motion estimation algorithm, summarizing the technology points and comparing their performance, we proposes a fast motion estimation algorithm. Methods Small Diamond Search Pattern (SDSP) and Hexagon (HSP and VSP) composition template are used to conduct template-matching search adaptively. Results The fast motion estimation algorithm classifies the motion type by making full use of the time and space correlation of the image sequences, and makes the initial search point prediction. Conclusion The results of the simulation experiment show that the improved algorithm could gather the search speed, shorten the search time and reduce the motion search points on the premise of the quality of reconstructed image.
出处
《宝鸡文理学院学报(自然科学版)》
CAS
2010年第2期61-64,共4页
Journal of Baoji University of Arts and Sciences(Natural Science Edition)
基金
宝鸡文理学院重点科研项目(ZK09165)
关键词
视频编码
运动估计
块匹配
搜索模式
video coding
motion estimation
block matching
search pattern