视频识别中基于簇的在线运动分割算法研究

仿射相机模型下,运动分割问题转化为子空间分离问题,处理这类问题的算法大多是离线算法,当假设不满足时性能很不理想.针对上述问题,提出一种在线运动分割算法,通过动态标签传输和簇分割进行运动分割.首先,根据固定数量的帧进行初始化,接着,通过在线策略更新轨迹相似性,最后,利用动态标签传输技术在帧间传输信息,对簇进行评估和归一化切分成本估计,实现动态的簇分割.基于基准数据集的仿真实验结果表明,算法的运行结果与离线算法相当.

Research on Online Motion Segmentation Algorithm Based on Clustering in Video Identification

Under the affine model,the motion segmentation problem becomes that of subspace separation.Due to this assumption,such methods are mainly off-line and exhibit poor performance when the assumption is not satisfied.In order to solve these problem we propose an approach that achieves online motion segmentation through dynamic label propagation and cluster splitting.Starting from an initialization computed over a mixed number of frames,we update the similarity between trajectories in an online fashion.After that,we propagate the label information from one frame to the next using dynamic label propagation,at the same time,evaluate each cluster and measure a normalized cut cost of splitting the cluster for dynamic cluster splitting.The performance of the proposed algorithm is evaluated on a benchmark dataset and achieves competitive performance while being online.