显著性能量目标跟踪轨迹修正算法

针对SiamMask在目标跟踪过程中,图像序列中出现运动模糊时跟踪机制无法捕获特征点而导致的跟踪漂移问题,提出一种显著性能量目标跟踪轨迹修正算法。该算法通过显著性能量特征判定是否发生跟踪漂移,利用轨迹预测算法修正发生漂移时的跟踪结果,解决运动模糊条件下跟踪漂移问题,进一步提高SiamMask算法跟踪精度。分别在OTB50和VOT2018数据集进行仿真测试,仿真结果表明该算法较SiamMask算法跟踪精度提高0.2%,有效修正跟踪漂移时的目标位置,适用于智能监控和自主驾驶系统等。

Saliency Energy Object Tracking Trajectory Correction Algorithm

The object tracking process of SiamMask algorithm has tracking drift issue under motion blur condition. When motion blur occurs in the image sequence, there are feature points which cannot be captured by tracking mechanism. To solve this issue, this paper presents a saliency energy object tracking trajectory correction algorithm (SEC) which can determine whether the tracking drift has occurred. And then the tracking result can be corrected to improve the tracking accuracy of SiamMask algorithm if there is a tracking drift. Simulation tests prove that, on the OTB50 and VOT2018 benchmark, the accuracy of SEC is 0.2% higher than the SiamMask algorithm. Our algorithm can correct the object position in the tracking drift and is more suitable for intelligent monitoring and autonomous driving system.