A Classification-aided Track Correlation Algorithm
LIU Wan-quan,XU Shi-you,SU Shao-ying,CHEN Zeng-ping
DOI: https://doi.org/10.3969/j.issn.1003-0530.2011.12.006
IF: 4.729
2011-01-01
Signal Processing
Abstract:A classification-aided track correlation algorithm is presented to improve the correlation performance using features attributes of the targets in a distributed multisensor information fusion system.First,track correlation problems using kinematic states classification information of the targets are transformed into corresponding hypothesis tests based on statistical mathematics theory.According to the distinguished factors which are set beforehand,the correlation matrices of kinematic states classification information are constructed using the statistics of kinematic states classification information of the targets.Then the fusion track correlation matrix is formed based on the logical and fusion rule.Finally,the minimum average statistical distance criterion is used to make a track correlation decision.By Monte Carlo method,the changing disciplines of correct correlation rate with the factors such as the number of correlation information point,the distance between targets in the same formation the coefficients of the process noise are simulated,using the track correlation algorithm based on target kinematic states the classification-aided track correlation algorithm. The simulation results show that,in the case of little number of correlation information points,dense targets distribution or large process noise,a better correlation performance can be obtained when the classification information is introduced into track correlation process.