基于块匹配的运动矢量可靠性分析

Motion vector reliability analysis based on block matching

为了改善目前对于运动矢量分析还较笼统的现状,获得对运动矢量场更为直观和定量的分析,运用贝叶斯定理,设计一种新型的基于运动矢量后验概率计算的矢量可靠性分析算法.利用贝叶斯理论中关于决策分析的思想,结合视频信号的特点和相关理论,计算运动矢量的后验概率,作为对矢量可靠性判断的依据,并据此优化运动补偿策略,提高去隔行效果.实验结果表明,用该法对运动矢量场进行分析,可得出对运动估计算法客观定量的分析结果,同时有助于矢量类去隔行算法的改进.运动矢量可靠性分析方法增强了块匹配运动估计算法的可分析性,有助提高运动补偿算法的鲁棒性和高质量地重建图像.

To improve the current situation that analysis of motion vector is still too general, a novel reliability analysis method of motion vector based on the posteriori probability of motion vectors from Bayes theory is described to gain more visual and quantitative analysis of motion vector field. Using the thinking of policy-making in Bayes theory and combining the characters and theories of video signal, as the basis of reliability judgement of motion vector, the posteriori probability of motion vectors is calculated to optimize motion compensation The results show that using this method to analyze tive judgement of motion estimation algorithm can strategy and enhance the effect of de-interlacing. the motion vector field an objective and quantitabe obtained and it is helpful to improve the effect of de-interlacing. With the reliability analysis method of motion vector, analysable extent of motion estimation algorithm based on block-matching is enhanced, the frame image can be reconstructed perfectly and robusticity of the motion compensation algorithm can be improved.