【摘要】 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.更多还原

【Abstract】 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.更多还原