An Effective Data Fusion and Track Prediction Approach for Multiple Sensors

Multiple sensor data fusion is a hot topic in the academic research. This paper developed an effective scheme to extract the flight trajectories from different sensors and searched their common characters by matching algorithm, which removed some abnormal points in each extracted trajectories and exploited cubic spline interpolation method to register the intersected parts of two trajectories which belongs to one target. Due to the accuracy of different observations from different sensors, the approach utilized by Least Square (LS) to estimate noise covariance for consequential processing, and then applied distributed Kalman filter to combine their measured trajectories to one target trajectory. Finally, the paper predicted target trajectory with prior knowledge and evaluated its accuracy via simulation, which showed the proposed approach had effectively integrated the multiple data and predicted the flight tracks.