CAJ | 학술논문

为有效预测运动矢量,降低运动估计复杂度,提出一种阈值选择算法.设定阈值以确定当前块是否为预测运动矢量块,对最佳运动矢量块做出判定,以提前终止运动估计.在JM10.2平台上与UMHexagonS算法进行对比测试,验证该算法的性能.实验结果表明,该算法缩短了约84%的编码时间,提高了编码效率,而视频质量无明显下降,满足实时应用的需求.
위유효예측운동시량,강저운동고계복잡도,제출일충역치선택산법.설정역치이학정당전괴시부위예측운동시량괴,대최가운동시량괴주출판정,이제전종지운동고계.재JM10.2평태상여UMHexagonS산법진행대비측시,험증해산법적성능.실험결과표명,해산법축단료약84%적편마시간,제고료편마효솔,이시빈질량무명현하강,만족실시응용적수구.
A threshold selection algorithm is presented to effectively predict vector and reduce the complexity of Motion Estimation(ME). Whether the current block is or not the Predicted Vector BIock(PVB) is decided by setting up a threshold, and some decisions on the optimization PVB are made in order to terminate the ME process as early as possible. Compared with the UMHexagonS algorithm, the performance of the algorithm isestified on the JM10.2 platform. Experimental results show that the coding time is approximately saved by 84% and the coding efficiency is greatly raised while the degradation of video quality is negligible for real-time application demand.