基于δ-GLMB自适应门限判定的多量测目标跟踪算法

An Multi-detection Target Tracking Algorithm Based on δ-GLMB Adaptive Threshold Determination

  • 摘要: 针对实际复杂交通场景中毫米波雷达能从单目标上获得多个量测导致广义标签多伯努利(δ-Generalized labelled multi-Bernoulli, δ-GLMB)滤波算法的多目标跟踪结果中出现单目标有多条轨迹等问题,提出了一种在δ-GLMB跟踪结果的基础上加入自适应门限判定的改进算法。首先,通过δ-GLMB滤波器对场景中目标进行跟踪,然后通过自适应门限判定方法实现目标多余轨迹点的删除和属于同目标的轨迹的标签统一。本文使用77GHz毫米波雷达对实际交通场景的监测数据进行了实验,结果表明本文提出的方法在目标个数估计准确率上有显著提高,对实际交通数据的鲁棒性更好。

     

    Abstract: In actual complex traffic scenarios, millimeter-wave radar can obtain multiple measurements from a single target, which can lead to multi-target tracking results of the δ-Generalized labelled multi-bernoulli (δ-GLMB) filtering algorithm may appear multiple trajectories for a single target. To address these issues, an improved algorithm with adaptive threshold determination based on δ-GLMB tracking results was proposed. Firstly, targets in the scene were tracked by a δ-GLMB filter. Then, using the adaptive threshold determination method to make the redundant track points of the target were deleted and the labels belonging to the same target were unified. In this paper, 77GHz millimeter-wave radar was used to conduct experiments on the monitoring data of actual traffic scenes. The results show that the method proposed by this paper has a significant improvement in the accuracy of target number estimation and is more robust to actual traffic data.

     

/

返回文章
返回