智慧交通中形变自适应的目标跟踪算法研究

Deformation-adaptive robust object tracking algorithm in intelligent transportation system

介绍了一种形变自适应更新策略,分析并构造了上下文峰值旁瓣比(Context Peak-to-Sidelobe Ratio,CPSR)特征,探究了自适应调整滤波器模板学习率。实验结果表明,在形变数据集上,DA-ECO算法优于ECO-HC算法,达到针对目标形变具有较强的鲁棒性。

This paper introduces a deformation-adaptive update strategy and constructs the Context Peak-to-Sidelobe Ratio(CPSR)feature.And explores self-adaptively learning rate for the filter template.The experimental results show that the DA-ECO algorithm is superior to the ECO-HC algorithm on the deformation dataset,and it is robust to the object deformation.