基于起点预测的单位十字快速运动估计算法

Unit-Rood Fast Motion Estimation Algorithm Based on Starting-Point Prediction

复杂且耗时的运动估计运算给实时视频编码系统的实现带来了困难.为提高视频编码的实时性,文中分析了运动矢量的分布特性和空间相关性,提出了一种基于起点预测的单位十字快速运动估计算法.该算法结合提前中止准则,通过块匹配绝对误差比较法来选择起始搜索点,然后采用单位十字搜索模式进行搜索.实验结果表明,在保持图像质量基本不变的情况下,该算法搜索速度是三步法的3～17倍,是菱形搜索法的2～9倍,是自适应十字搜索法的1.19～4.42倍.该算法计算量小,实时性强,易于硬件实现,在小运动序列运动估计方面具有明显优势.

Complex and time-consuming computation in motion estimation poses a great challenge to the implementation of real-time video coding. In order to improve the real-time performance in motion estimation, the distribution characteristics and spatial correlation of the motion vectors are analyzed, and a unit-rood fast motion estimation algorithm based on the starting-point prediction is proposed. In the proposed algorithm, the early stop criterion is adopted to predict the starting-point of search by comparing the sum of absolute difference values of the candidate points, and a unit-rood pattern is used to perform the search. Experimental results show that the proposed algorithm, which is of a searching speed of 3 - 17 times that of the Three-Step Search algorithm, 2 - 9 times that of the Diamond Search algorithm and 1.19 ~ 4. 42 times that of the Adaptive Rood Pattern Search algorithm, maintains the picture quality well. Moreover, as compared with other motion estimation algorithms, the proposed algorithm is easy to implement and is of less computational complexity, better real-time performance, and performs better in terms of small motion video sequence.