论文部分内容阅读
提出了一种在H.264压缩域下进行运动对象分割的新算法。算法主要利用H.264码流中的运动矢量信息来进行对象分割,为了提高运动矢量信息的鲁棒性,首先利用I帧中的帧内预测模式和预测残差能量进行区域划分;在P帧中,利用帧间预测残差能量来更新区域划分结果,对部分区域的运动矢量进行归零化处理。再根据P帧中的分块模式,采用不同的滤波器对运动矢量进行滤波;最后利用滤波后的运动矢量信息建立对应的Gibbs势能函数,采用迭代条件模式方法求解最大后验概率,得到可靠的运动对象标记。实验结果表明,该运动对象分割算法可以获得有效并可靠的分割结果。
A new algorithm for segmenting moving objects in H.264 compression domain is proposed. In order to improve the robustness of the motion vector information, the algorithm first uses the intra prediction mode in I-frame and the prediction residual energy to segment the region. In P-frame , The inter-frame prediction residual energy is used to update the area division result, and the motion vector in a part of areas is zeroed-out. Secondly, the motion vector is filtered by different filters according to the block mode in P-frame. Finally, the Gibbs potential energy function is established by using the filtered motion vector information, and the maximum posteriori probability is obtained by the iterative condition mode method to obtain a reliable Movement object mark. Experimental results show that the segmentation algorithm can obtain effective and reliable segmentation results.