自适应上下文感知相关滤波类目标跟踪算法

Algorithm for tracking adaptive context-aware correlation filter targets

由于传统的相关滤波目标跟踪算法余弦窗和搜索区域的限制,在复杂场景下容易产生跟踪漂移,上下文感知算法提出一个框架允许将上下文纳入相关滤波器,但其没有计算上下文信息对目标的干扰程度而直接采用相同的抑制权重,因此无法对干扰程度不同的上下文信息自适应赋予不同程度的抑制。基于此,提出一种上下文抑制权重自适应的相关滤波类目标跟踪算法。首先,将目标周围的背景信息学习到滤波器中,增强滤波器模板对于目标和上下文背景信息的分类能力,同时引入自适应权重系数向量;其次,提出一个上下文信息干扰系数公式,用于定量评估目标上下文信息对于目标的干扰程度;再次,依据所提出的公式分别计算出上下文信息的干扰程度之后,将其与自适应权重系数向量匹配,从而实现对目标干扰程度越大的上下文信息,抑制的程度越大;最后,基于OTB100数据集验证了该算法的有效性。实验结果表明,这种算法的成功率和精确度较其基准算法分别提升了约5.7%和4.3%,具有较强的鲁棒性。

In the traditional correlation filter object tracking algorithm,due to the limitation of the cosine window and the search area,a tracking drift can easily occur in complex scenes.A framework is proposed in the context-aware algorithm to allow global contextual information to be incorporated into the correlation filter tracker,but the same suppression weight is directly used instead of calculating the interference degree of the context information to the target and it cannot adapt to giving background information with different degrees of interference,on the basis of which we propose a context suppression weight adaptive correlation filter target tracking algorithm.First,the background information arpond the target is learned into the filter,enhancing the classification ability of the filter template for target and context background information,and the adaptive weight coefficient vector is introduced.Second,a formula for the context information interference coefficient is proposed to quantitatively evaluate the interference degree of the context information to the target.Third,according to the proposed formula,the interference degree of context information is calculated,and then it is matched with the adaptive weight coefficient vector,so as to lead to the effect that,the greater the interference degree of the context information to the target,the greater the suppression degree.Finally,we rely on the OTB100 dataset to verify the effectiveness of the algorithm in this paper,with experimental results showing that the success rate and accuracy of the algorithm in this paper are improved by 5.7%and 4.3%,respectively compared with the benchmark algorithm,and that it has strong robustness.