基于双模型融合的自适应目标跟踪算法

针对目标跟踪过程中的光照变化、背景混乱和目标形变等问题,提出一种背景抑制的HS直方图和核相关滤波双模型融合的自适应跟踪算法。首先引入非线性核相关滤波跟踪模型；其次提出背景抑制的HS颜色直方图跟踪模型,通过分离亮度分量以减小光照干扰,并采用背景加权突出目标信息；然后提出一种自适应融合策略,根据目标与背景的HS特征相似度来动态调整两个模型融合权重,以降低背景混乱和目标姿态变化的影响；最后针对目标尺度变化问题,采用尺度金字塔估计策略进行解决。实验表明,与现有算法相比,提出的算法能更好降低光照、背景混乱等复杂因素干扰,鲁棒性更强。

Adaptive target tracking algorithm based on the fusion of two models

An adaptive tracking algorithm using the fusion of background suppressed HS histogram model and correlation filter model is proposed to deal with variable kinds of complicated tracking problems, such as illumination variation, background clutters, object deformation and so on. Firstly, a non-linear kernelized correlation filter tracking model is introduced. Secondly, a background suppressed HS color histogram tracking model is proposed. The luminance component is separated to reduce the interference of illumination and a background weighted method is used to highlight object information. Furthermore, an adaptive fusion strategy is proposed. The HS feature similarity between the object and each background patch is computed to adjust the fusion weight of two models dynamically. In this way, the proposed algorithm can reduce the influence of background clusters and object pose variation. Finally, a scale pyramid estimation strategy is used to handle large scale variations. Experimental results demonstrate that, compared with other trackers, the proposed approach performs better in regard of complex interference factors, such as illumination variation and background clusters, and has stronger robustness.