摘要
MPEG- 4是一个基于对象的视频标准 ,其运动估计方法不同于其他视频标准。本文提出一种基于对象的菱形搜索运动估计算法 ,利用图像分割信息和运动矢量的相关性 ,减少了得到运动矢量所需要的平均搜索点数。实验结果表明 ,在降低计算复杂度的同时 ,本算法保持了较好的图像质量。
MPEG-4 is an object-based video standard, the motion estimation method of MPEG-4 is different from other video standards. This paper proposes an object-based motion estimation algorithm using diamond search. Based on the segmentation information of the picture and the correlation of motion vectors, the proposed algorithm can reduce the average search points in getting motion vectors. Experimental results show that the algorithm is computationally efficient while keeping better image quality.
出处
《数据采集与处理》
CSCD
2001年第1期110-114,共5页
Journal of Data Acquisition and Processing
