针对尺度变化、目标形变、背景混乱及相似等导致的相关滤波跟踪算法模型漂移的问题，本文提出了一种基于背景抑制和前景抗干扰策略的多尺度相关滤波跟踪算法。前者应用自适应高斯窗和颜色概率模型，以解决背景混乱及相似问题；后者采用自适应密集采样和颜色概率模型，以抑制尺度变化和目标形变等干扰。最后，构建一维尺度滤波器实现对目标尺度的精确估计。在 OTB-50数据集下的对比实验表明，本文算法取得了 79.2%的精确度和 65.5%的成功率，优于现有的主流相关滤波跟踪算法，在 11种常见的干扰下性能亦为最优，展示出较高的跟踪精度和较强的鲁棒性。

To deal with the model drift of the correlation filter caused by scale changes, target deformation, background confusion and similarity, a multi-scale correlation filter tracking algorithm based on suppression for background and anti-disturbance for foreground is presented in this paper. The former applied the self-adaption Gaussian window and color probability model to solve the problem of background confusion and similarity; the latter used self-adaption dense sampling and the color probability model to deal with scale change and target deformation. Finally, a one-dimensional scale filter for precise estimation of target’s scale was constructed. The experimental results on OTB-50 datasets demonstrate that this algorithm attains a precision of 79.2% and a success rate of approximately 65.5%. It outperforms the mainstream correlation filter trackers in the case of 11 common disturbance factors, demonstrating high tracking accuracy and strong robustness.