基于粒子滤波的活动轮廓模型视频跟踪(英文)

跟踪变形目标包括假定目标的全局运动和时间方程的局部变形。粒子滤波算法依靠参数化法选择,但是不能处理曲线的拓扑变化。活动轮廓模型是独立参数化,可以比较好地适应拓扑变化。基于上面两个方法的局限性,将上述两种方法相结合,运用于变形目标的跟踪。实验结果证明,该方法有很强的鲁棒性,对部分遮挡的物体有很强的适应性,证明了该方法的有效性。

Geometric Active Contours Using Particle Filtering

Tracking deforming objects involves estimating the global motion of the object and its local deformations as a function of time. Tracking algorithms using particle filtering is dependent on the chosen parametrization and cannot process changes in curve topology. Geometric active contours which is paramaetrization independent and allow for changes in topology is adopted. A particle filtering algorithm in the geometric active contour framework is formulated that can be used for tracking moving and deforming objects. The experimental results show that the algorithm is computationally efficient and robust to changes in topology, patial occlusion and interactions of non-rigid objects.