基于高置信度更新策略的高速相关滤波跟踪算法

High-Speed Correlation Filter Tracking Algorithm Based on High-Confidence Updating Strategy

为了满足在线目标跟踪算法的实时性需求并提高算法的稳健性,提出一种基于高置信度更新策略的相关滤波跟踪算法。在目标区域提取、融合多特征,以构建稳健的外观表达,并利用投影矩阵对特征进行降维,以提高算法的运行效率;通过相关滤波器寻找最大响应值,从而快速定位目标;利用最大响应值和平均峰值相关能量指标,设计了一种高置信度更新策略。结果表明:所提算法在大规模公开数据集上取得了较高的跟踪精度和成功率,平均跟踪速度达到122.3 frame/s。

To satisfy the real-time requirements of the online object tracking algorithm and improve the robustness of the algorithm, we propose a correlation filter-based tracking algorithm with high-confidence updating strategy. Multi-features are extracted and integrated in the target region to construct robust appearance representation, and the projection matrix for dimension reduction of features is used to improve the operational efficiency of the algorithm. The correlation filter is used to localize the target at a high speed via the maximum response value. Two indicators of maximum response value and average peak-to-correlation energy are utilized to design a high-confidence updating strategy. The results show that the proposed algorithm achieves high tracking precision and success rate on large-scale public datasets while running at 122.3 frame/s on average.