【作者】 赵晋； 张建秋； 胡波；

【Author】 Zhao Jin, Zhang Jian-qiu, Hu Bo(Dept. of E.E., Fudan University, Shanghai 200433

【机构】 复旦大学电子工程系；

【摘要】 本文提出一种新的目标跟踪算法,将鲁棒外观滤波与粒子滤波器相结合,从而实现稳定的目标跟踪。在传统的跟踪算法中,外观模型是固定的或迅速变化的。在本文中,模板的更新将通过对每个象素的外观特征进行鲁棒卡尔曼滤波来实现。这样处理后的模板,在目标被局部遮挡的情况下,也可以进行准确的跟踪。该方法需要设置的参数很少,在运算量上可以满足实时目标跟踪的要求。更多还原

【Abstract】 A new approach that incorporates robust appearance filter with a particle filter to realize robust object tracking is presented in this paper. In conventional tracking algorithms, the appearance model is either fixed or rapidly changing. Here, appearance features are smoothed temporally by robust Kalman filters to update the template, one to each pixel. The resistance of the resulting template to partial occlusions enables the accurate detection. The proposed method has only a few parameters and is computationally fast enough to track object in real time.更多还原