基于联合运动补偿的边信息预测算法

A Side Information Prediction Algorithm Based on Joint Motion Compensation

摘要:

传统块运动补偿算法仅利用单一运动向量场,该运动向量场中总会存在一定的运动异常,而过多的运动异常会严重衰退内插帧质量。为了解决该问题,提出了一种基于联合运动补偿的边信息内插算法,该算法使用双向运动估计计算出内插帧的运动向量场,接着采用当前块和其八个邻域块的运动向量联合预测出目标块。由于运动向量场的局部平滑特性会使八邻域块的运动向量十分接近当前内插块的真实运动向量,实验结果表明提出算法比传统算法有更好的容错性能。

Abstract:

Traditional block motion compensations only use single motion vector field, There are some movement abnormalities in the motion vector field, The excessive movement abnormalities declines quality of interpolated frame. In order to solve this problem, a side information prediction algorithm based on joint motion compensation is proposed. The method uses bidirectional motion estimation to calculate motion vector field of the interpolated frame, then it uses motion vector of the current block and its eight neighboring block to joint predict the target block. The motion vector of the eight neighboring block very closes to the true motion vector of the current interpolated block, because of local smoothing feature of motion vector field. The results show the proposed algorithm improves the fault-tolerance of motion compensation.