A Track to Track Association Algorithm Based on Weighted State Correlation Similarity

Abstract

In multi-sensor systems, track association plays a critical role to ensure an accurate multi-target tracking. In this study, we propose a novel statistical method based on temporal state correlation similarity. In this method, a hybrid distance metric is derived from the correlation coefficients of the covariance matrix obtained from the sequential states of individual tracks and the distances between different target states. Contrary to many association algorithms that perform association in every single scan, the proposed method processes the track states as blocks in a given time period. The effectiveness of the proposed method under unbiased sensor measurements is illustrated by various three-dimensional multi-target tracking simulation scenarios where target density and the sensor noise level significantly varies.