Research On Target Localization Method Based On Multi-Sensor Information Fusion

Detection and localization of enemy targets is a very important mission in war and various sensor localization technologies have been applied in the modern military field.For singlesensor system,detection range and accuracy are limited,however,by utilizing information from multiple sensors,the overall detection range can be effectively extended and the accuracy of localization results can be improved.Problems to be solved in the process of multi-sensor localization include accurate location estimation,effective data association and multi-sensor information fusion.Multi-sensor localization algorithms are studied in depth and research contents are as follows.The theory and algorithms of localization based on seismic signal and azimuth angle are studied.For TDOA localization algorithms,Chan algorithm can hardly achieve high accuracy while Tylor algorithm needs initial estimation and is not easy to converge.To solve these problems,Chan-Tylor joint localization algorithm is employed.For AOA localization problems,the global least squares estimation algorithm is used and the method is much more effective compared with traditional least squares estimation algorithm.Aiming at solving the problem of data association in multi-target scenarios,multidimensional assignment association algorithm based on probability distribution of targets is proposed.The algorithm pre-allocates sensor measurements according to the probability distribution of targets so as to improve the computational efficiency.Data association problem in localization based on visual azimuth is solved by visual feature matching.Taking advantage of fast matching of ORB features,data association algorithm based on ORB feature matching is used.To solve the problem that the association effect deteriorates with the increase of target distance,an improved algorithm based on multi-dimensional assignment of target azimuth is proposed.In the aspect of multi-sensor localization fusion,the fusion processing is respectively analyzed by utilizing sensor measurements and localization results.The mathematical model of TDOA-AOA joint localization is established in this paper and the weighted least squares algorithm to solve the localization problem is derived based on the model.In addition,fusion algorithm based on the best linear unbiased estimation is employed to fuse the localization results.The covariance matrix of the localization error is difficult to calculate for the real target positions are unknown in practice.To solve the problem,on-line calculation method of covariance matrix is presented and fusion localization algorithm based on best linear unbiased estimation is realized.On the basis of theoretical research,the multi-sensor joint localization system based on seismic signal and azimuth angle is designed and graphical interface program of the system is developed.With actual data collected from sensors,experiments for each processing part of the system are carried out and the effectiveness of proposed algorithms is verified.