Research On Multi-sensor Estimation And Fusion Problem For Mixed Target

Mixed target is made up of point targets,group targets and extended targets.The point target can produce at most one measurement,the extended target can produce multiple measurements,the group target can produce multiple measurements and its sub-objects have a certain dependency relationship.Usually,multiple sensors are used to track and estimate mixed targets,so as to obtain multiple sets of estimated tracks.At this time,the fusion of multiple sets of estimated trajectories becomes particularly important.The main content of this thesis is as follows:(1)Multi-sensor estimation fusion algorithm based on OSPA⁽²⁾distance and federated filter.Under multi-sensor condition,the target tracking process generally produces multiple sets of estimated trajectories,and the correlation of multiple sets of estimated trajectories becomes a key issue for target estimation fusion.This thesis uses OSPA⁽²⁾distance to calculate the distance between arbitrary local track sets,,the OSPA⁽²⁾distance of the different point target track is greater than the set threshold,and the OSPA⁽²⁾distance of the same point target track is the smallest,so as to judge the track sets belonging to the same point target,and then a federated filter fusion algorithm is used to fuse the track sets belonging to the same point target.(2)A mixed target estimation fusion algorithm based on Gibbs-GLMB and federated filter.Group targets have multiple measurements,and the distances between their sub-targets are relatively close,which brings difficulties to the estimation and fusion of mixed targets.Firstly,the adjacency matrix in graph theory is used to dynamically model the distinguishable group target,and introduce the concept of deformation,the finite mixture model method is used to dynamically model the extended target.Then the Gibbs-GLMB algorithm is used to estimate the state and number of mixed targets.Finally,a dynamic detection federated filter fusion algorithm is proposed to fuse the mixed target state estimation.(3)A mixed target estimation fusion algorithm based on equivalent measurement.Indistinguishable group targets and extended targets are difficult to distinguish in terms of structural characteristics,which complicates the estimation and fusion of mixed targets including point targets,extended targets and indistinguishable group targets.Firstly,the graph theory model,centroid and upper and lower boundary points of the indistinguishable deformation group targets are given.Then the shape parameters are defined,the centroid and the upper and lower boundary points are estimated by the equivalent measurement method.Finally,the federated filter fusion algorithm of dynamic detection is used to fuse point targets,extended targets and the centroids of indistinguishable deformation group targets,as well as their upper and lower boundary points.