Research On Midcourse Targets Correlation Across Dissimilar Sensors Based On Multi-feature Data Fusion

Multi-sensor information fusion technology as a developing and crossing discipline spring from the requirement of multi-source information processing by modern warfare. nowadays,Multi-sensor information fusion has become one of the most important way of processing various kinds of information and data.As a result,synthesizing those kinds of information to achieve the best reasoning and decision has been a hotspot subject in information processing.Due to its robustness,broaded sights,and superior complementary advantage of measurements,more and more researches of combining multi-sensor information fusion and multi-target tracking has been taken.In distributed multi-sensor system,one of the key issues is to solve the data correlating problem,which is also the primacy of its application in multi-target tracking.Basic theory and processing architecture of information fusion are firstly introduced.Then some famous data correlating algorithms are concluded.Afterwards,a correlating algothm is formally raised to attend to radar and IR sensor measurements correlating,which is based on multi-feature fusion method.The targets correlating method based on multi-feature fusion is put forward considering the complex multi-target correlating situation across distributed sensors.By analyzing the shared features acquired from measuring sensors,the correlation measurement is then fused on those features with certain rules,which becomes the evidence of correlating decision.After a rough decision is made,a strategy is taken to optimizing it,thus to stress the features used and improve the result.Finally,a series of simulations are taken to test the method,and comes the result that,under appropriate circumstances,the method can make effective target correlation across dissimilar sensors,which makes a foundation of its application in multi-target tracking area.Considering the task of multi-sensor multi-target tracking,the correlation problem has been taken further research.As we know,in Kalman filter,the target prediction and estimation parts both need targets’ history data to revise targets’ states,and reduce the tracking error.As a conclusion,both targets correlating across dissimilar sensors and history data correlating are required in multi-target tracking across distributed sensors.Based on the first part result of target correlation method,a revised one is introduced to make history data correltion based on multi-feature fusion.Simulations is taken as well,and shows that the correlation method also performs convincing in multi-traget tracking,and meets the expectation.