论文部分内容阅读
为了减小快速运动估计算法的搜索速度和提高运动补偿的准确性,本文提出了一种线性预测搜索方向的快速运动估计算法。该算法在正方形模板块失真匹配的基础上,确定中心点到最小误差点为搜索方向,在此方向上不断延伸一个点进行块失真匹配,直至下一个点的块失真大于当前点的块失真。以当前点作为最小误差点,重复以上操作,直至找到最优点结束搜索。实验表明本算法在保证图像质量的情况下搜索速度比DS搜索算法提高1.5倍左右。
In order to reduce the search speed of fast motion estimation algorithm and improve the accuracy of motion compensation, a fast motion estimation algorithm for linear prediction search direction is proposed in this paper. Based on the distortion matching of the square template block, the algorithm determines the minimum error point from the center to the search direction and extends a point in the direction to match the block distortion until the block distortion at the next point is greater than the block distortion at the current point. With the current point as the minimum error point, repeat the above operation, until finding the best point to end the search. Experiments show that the proposed algorithm can improve the search speed by about 1.5 times than the DS search algorithm while guaranteeing the image quality.