基于最小匹配误差方向预测的快速半像素运动估计

为了减小半像素搜索的计算量,本文提出了一个基于最小匹配误差方向预测的快速半像素运动估计算法.本文提出的算法利用亚像素搜索窗内的匹配误差单峰曲面的特性来预测半像素搜索区域中最小匹配误差方向,从而避免了大量不必要的匹配运算量.实验结果表明,对于各种不同运动程度和空间细节的视频序列,本文提出的算法在保证和半像素全搜索法相同图像质量的同时,平均节省73%的计算量,很适合实时应用.

A Fast Half-Pixel Motion Vector Search Algorithm

A minimal matching error direction prediction-based fast half-pixel motion vector search algorithm is proposed in this paper in order to reduce computation load of half-pixel search. Based on the single peak characteristic of half-pixel error matching function inside search grid, the algorithm predicts minimal matching error direction with the help of SAD values comparison results of four integer-pixel points around integer-pixel motion vector, so that a great number of unnecessary computations for matching search can be avoided. The experimental results reveal that, to all kinds of video sequences, the proposed algorithm can obtain almost the same video quality as that of the half-pixel full search algorithm with reduced average 73% computation cost, which is very useful for real-time applications.