背景加权的多特征融合目标跟踪算法

针对单一特征目标跟踪导致多数跟踪算法鲁棒性差的原因,提出一种背景加权的多特征融合目标跟踪算法。在跟踪过程中对目标模型进行背景加权,同时利用空间直方图提取目标颜色的空间分布信息。在粒子滤波框架下将背景加权直方图和空间直方图相结合,并且引入特征不确定性度量,自适应调整不同特征对跟踪结果的贡献,有效地提高了算法的鲁棒性。实验结果表明:与传统融合算法相比,提出的算法具有更强的鲁棒性,同时提高了跟踪精度。该算法在目标表示和跟踪性能上都有很大的提高。

Fusing Multi-feature Object Tracking Algorithm Based on Background-weighting

Object tracking using single feature often leads to a poor robustness. In this paper, an object tracking algorithm using multi-feature fusion based on background-weighting is presented. In order to enhance the important features, the target model is background weighting while tracking to get an accurate color model of the object. Meanwhile, special histogram is used to obtain spatial layout of these colors for the target. These features are rationally fused in the framework of Particle filter. Uncertainty measurement method is then introduced into features fusion to adjust the relative contributions of different features adaptively, and the robustness of the algorithm is significantly enhanced. Experimental results indicate that the proposed algorithm is more robust and has good performance in complex scene. The use of the algorithm improves the accuracy of tracking and can track objects effectively even with similar color disturbance.