【摘要】 本文提出了一种新的在MPEG-4编码过程中提高实时性的算法。算法采用双重筛选机制,充分利用了运动估计的结果提取运动对象。首先在运动估计过程中加入背景优先准则进行第一重筛选;然后根据运动估计后宏块的最小平均绝对差值分利用区域自适应门限进行第二重筛选,从而获得运动区域掩模;最后对掩模进行形态滤波和填充消除噪声点,最终得到运动区域。算法分别在硬件和软件平台上进行测试,结果表明提出的算法得到了理想的分割效果,较传统的在编码前进行运动对象提取的方法实时性大大提高了。更多还原

【Abstract】 New algorithms,which aims to enhance the real time capability by using diplex-filtration mechanism and taking full advantage of the result of motion estimation in order to rapidly and ideally extract the moving region in the process of MPFG-4 coding,is proposed.Firstly,add some relevant rules to the process of motion estimation for the first filtration.Then carry through the self-adapting filtration according to minimum SAD,and the mask of the moving region is obtained.Lastly,after the morphology filter and removing noises,the accurate matrix of the motion regions is finally obtained;thereby the moving video object is also gained.Experimental results show that the algorithm mentioned above can lead to the satisfying segmentation effects and save tremendous amount of time compared to the traditional method of moving video object extraction before coding.更多还原