In object segmentation, motion features are utilized to improve segmentation accuracy.
而视频对象分割充分利用了运动特征来实现更有效和精确的分割,运动特征体现了视频的时域相关性。
参考来源 - 基于内容的视频应用中运动对象分割与运动估计技术的研究·2,447,543篇论文数据,部分数据来源于NoteExpress
We can acquire not only transfer motion features but also rotation motion features.
所提出的方法既能获取镜头的平移运动特征,也能获取镜头的旋转运动特征。
By modify motion model, targets with different motion features could be detected. Experiments show that the proposed method is feasible and robust.
通过调整运动模型,可以应用于不同运动特征的目标检测问题。实验结果证明了本文方法的有效性和鲁棒性。
This method first extracted the global motion and local motion features in each frame, and video key frame was extracted by Particle Swarm Optimization (PSO) adaptively.
该方法首先提取出视频中每帧的全局运动和局部运动特征,然后通过粒子群算法自适应地提取视频关键帧。
应用推荐