基于人类视觉智能和粒子滤波的鲁棒目标跟踪算法

与计算机视觉目标跟踪算法遇到的各种困难和瓶颈不同的是,鲁棒视觉目标跟踪仅仅是人类视觉系统的一项基本功能.为此,基于自顶向下的视觉注意机制,构建了一种相应的计算模型,提出一种利用检测视觉注意力焦点区域的方法来模拟人类视觉系统的全局搜索过程,并用粒子滤波跟踪来模拟人类视觉系统的局部跟踪过程的目标跟踪算法.多组对比实验结果表明了所提出的算法在视觉目标跟踪中的优越性.

Human visual intelligence and particle filter based robust object tracking algorithm

Computer visual object tracking algorithms suffer from many difficulties and bottlenecks. However, robust visual
object tracking is just a basic capability of human visual system(HVS). Therefore, a computational model simulating topdown
visual attention mechanism is constructed, and an object tracking algorithm is proposed by detecting focus area of
visual attention to imitate the global searching process of HVS and applying particle filter tracking to emulate the local
tracking process of HVS. Comparative experimental results show the outperformance of the proposed algorithms in visual
object tracking.