.png){kind=icon}
.png){kind=icon}
Return
Improved strong tracking adaptive interactive multi model tracking algorithm for highly maneuverable targets
.png){kind=icon}
PRE
AI
DOI：10.1016/j.phycom.2025.102915.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
To solve the problem of slow model switching and large filter prediction errors leading to low tracking accuracy in interactive multi model target tracking algorithms when tracking highly maneuverable targets, an improved strong tracking adaptive interactive multi model tracking algorithm is proposed. A sliding window is employed to calculate correction factors from historical residuals, enabling real-time adjustment of the transition probability matrix using posterior information. Furthermore, a modified strong tracking fading factor is integrated into the Unscented Kalman Filter, which dynamically adjusts the covariance matrix to improve estimation accuracy. The simulation results demonstrate that, compared with the conventional interactive multiple model algorithm, the proposed method increases the model matching probability by 11.14%, while reducing the position RMSE by 68.89% and the velocity RMSE by 39.80%. The proposed algorithm accelerates motion model switching and reduces filter prediction errors when tracking highly maneuverable targets, thereby enhancing overall tracking performance.
Keywords:
Target tracking
Sliding window
Historical residual
Correction factor
Strong tracking filter