基于贪心策略的多目标跟踪数据关联算法

Multi-target Tracking Data Association Algorithm Based on Greedy Strategy

摘要:

摘 要：针对多目标跟踪中数据关联问题，提出一种新的数据关联方法，该算法先计算航迹和点迹的欧式距离以及其状态向量的在1范数下的距离，并将两者的和作为关联测度，构建关联概率矩阵.根据关联概率矩阵，对每条航迹都找到最适合（关联概率最大）的点迹，若点迹只是一条航迹的候选点迹则予以更新，若点迹是多条航迹的候选点迹，则选择其中概率最高的一条航迹予以更新.蒙特卡罗仿真表明，该算法在最大程度上保证了对每条航迹更新的点迹尽量是当前所有点迹中最优的点.

Abstract:

Abstract:In this paper,a new association method is proposed to tackle the data association problem of multi-target tracking.In this algorithm, building the associative matrix with the Euclidean distance and the 1-Norm of state vector between tracks and points firstly.And using the associative matrix find the most suitable(Maximum matching success rate)points for every track. If the points just marked by one track, update this track directly; if the points marked by many tracks, choose the track with highest probability to update. Monte-Carlo Simulation experiments show that this algorithm guarantees the updating points for every tracks are the best points among all present points.