Research Of Target Tracking Algorithm Based On Multi-source Sensor Information Fusion

In recent years,the multisource sensor fusion technology is one of the hottest topics in the field of information science.With high precision and low cost,it is widely used in military and civilian areas,and target tracking is one of the important research directions.In this thesis,based on an photoelectric theodolite tracking system,the precision and stability of target tracking are improved by improving the multi-sensor information fusion algorithm.First,the basic theories and several common algorithms of the multi-sensor information fusion are briefly described and the target tracking technology based on coordinate transformation is introduced.Second,because poor measurement and target motion mutation in the non-linear Gaussian system can cause tracking performance degradation,the target tracking algorithm of the multi-sensor information fusion based on STACKF is proposed.According to the CKF algorithm and the innovation covariance matching principle,this thesis introduces the noise factor and combines with the strong tracking filter theory to establish the robust STACKF algorithm,and applied it,through an one-step predictive fusion method,into the multi-sensor information fusion.Though the simulation experiments,the algorithm shows that it ensures the adaptive tracking ability,improve the tracking performance and fusion precision.Finally,a data fusion tracking algorithm based on STCPF is proposed to solve the problem of particle degradation in non-linear non-Gaussian systems,as well as the increaseing of errors caused by the sudden change of target motion and the divergence of filter tracking.In the framework of PF algorithm,the STCPF filter tracking algorithm is established by selecting the appropriate importance density function,and the data of multiple sensors are fused by on-line adaptive weighted fusion.The simulation results show that the improved algorithm can effectively improve the target's estimation precision and tracking performance under non-Gaussian noise,and reduce the computation errors.