相关滤波-粒子滤波的长时目标跟踪算法

Long-Term Target Tracking Algorithm Based on Correlated Filtering-Particle Filtering

针对大多数目标跟踪算法在长时跟踪过程中目标遮挡、形变等干扰属性导致不能有效跟踪的问题,提出一种基于相关滤波-粒子滤波协作的长时目标跟踪算法(CFPE).首先相关滤波器中目标特征表达采用融合CN特征和HOG特征,增强在复杂情况下的目标描述能力;然后通过平均峰值相关能量(APCE)指标和最大响应值对当前目标位置做出判断并决定滤波器模板是否更新,当判断跟踪失败时,通过粒子滤波重新检测目标位置;最后在OTB100和UAV123视频集中和近年来优秀的跟踪算法中进行试验分析,试验结果表明:CFPE具有稳定长时跟踪和实时性的优势,并且在背景变化、形变和遮挡方面优于大部分算法.

In order to solve the problem that most target tracking algorithms cannot track effectively due to interference properties such as target occlusion and deformation during long-term tracking,a long-term target tracking algorithm(CFPE)based on correlation filtering-particle filtering cooperation is proposed.First,the target feature expression in the correlation filter uses the fusion of CN and HOG features to enhance the ability to describe the target in complex situations;then the Average Peak Correlation Energy(APCE)index and the maximum response value are used to judge the current target position and determine the filtering whether the updater template is updated.When the tracking failure is detected,the target position is re-detected by particle filtering.Finally experimental analysis are made in the OTB100 and UAV123 video sets and the excellent tracking algorithms of recent years.The test results show that CFPE has stable long-term tracking and real-time and it is better than most algorithms in background changes,deformation and occlusion.