【摘要】 本文应用率失真理论在固定码率的前提下对可变尺寸块运动估计过程进行R D优化 ,获得了最佳预测块划分和最小运动补偿误差 ,并提出了基于树形结构的运动矢量编码方案。R D优化问题的求解通过引入失真度 -码率梯度 ,将R D优化与分层运动估计结合在一起 ,克服了G BFOS算法需要对各个分块层次进行运动估计及所得局部最优解严格位于R D曲线凸包络线上的弊病。软件模拟结果显示本算法大大降低了G BFOS优化算法的计算量 ,且运动补偿PSNR值高于G BFOS算法结果。更多还原

【Abstract】 The rate distortion theory is applied to optimize the variable size block motion estimation,the purpose of which is to obtain the best block segmentation and the least compensation error given a bit rate constraint A scheme based on a tree structure is presented to encode the motion vector The distortion rate gradient is used as a heuristic information in the R D optimization in this paper Our approach avoids the estimation of the whole hierarchical motion field by the combination of the motion estimation and rate distortion optimization,and achieves better sub optimal solutions by not restricting ourselves to operating points on the convex hull,which are two major problems of existing G BFOS algorithm Simulation results show that the proposed algorithm can greatly reduce the complexity of G BFOS algorithm with increasing PSNR value of motion compensation更多还原