自适应尺度特征融合与模型更新的跟踪算法

在核相关滤波器跟踪算法中，为了减少背景相似物等杂波对跟踪器的干扰，以及解决不同跟踪结果置信度下的模型更新问题，提出了自适应尺度特征融合与模型更新的跟踪算法。通过多特征融合和尺度变化策略改进了多特征的尺度核相关滤波器，使用多峰检测对响应图的整体振荡程度进行判断，再对峰值进行跟踪结果置信度评估；在遮挡、形变等跟踪结果置信度低的情况下及时停止模型更新，在高置信度模型更新时，引入初始模型进行对齐操作，减少模型的更新误差，抑制模型漂移。比较核相关滤波器算法，本算法准确度较高，且在目标尺度变化、遮挡和形变时稳定性更好。在OTB-50数据集上的实验结果表明，该算法在精度和成功率上都比核相关滤波器算法表现更优。

Self-adaptive scale feature fusion and model update tracking algorithm

In the kernel correlation filter tracking algorithm, in order to reduce the interference of background similarity and other clutter to the tracker, and to solve the problem of model updating under the different confidence degree of tracking results, this paper proposed a self-adaptive scale feature fusion and model update(SFMU) tracking algorithm. Through multi-feature fusion and scale variation strategy to improve the multi-feature scale kernel correlation filter, it used multi-peak detection judged the overall oscillation degree of response map, then evaluated the confidence degree of the tracking result. The algorithm stopped updating model timely in the case of low confidence of the tracking results such as occlusion and deformation. In the high confidence model update, the algorithm introduced the initial model to the alignment operation to suppress model drift. Therefore, compared with the kernel correlation filter algorithm, this algorithm is more accuracy, and the stability is better in the target of scale variation, occlusion and deformation. The experimental results on the OTB-50 dataset show that the precision and success rate of the proposed algorithm are better than those of kernel correlation filter algorithm.