Research On Multi-target Trajectory Tracking Technology Based On Multi-sensor

With the continuous improvement of computer hardware performance,data processing technology,software simulation ability and algorithm accuracy,research on multi-target trajectory tracking technology based on multi-sensor has been widely used in various fields.Comparing with the single sensor or sensor cluster acquisition of fragment information analysis technology,the multi-sensor cooperation is more stronger and accuracy.It has important theoretical and practical significance to realize all-time,all-round and accurate identification of the target through information complementation.In view of the above requirements,this paper discusses and studies the key issues of multi-sensor and multi-target tracking in the following aspects:(1)In terms of multi-sensor cooperative networking detection.Firstly,the architecture of the whole sensor network is analyzed and studied,then the collaborative control methods of multiple sensors are compared and discussed,and the classic traditional DCSP and organization design method are studied.On this basis,the dynamic network coordination technology in complex environment and multi climate state is determined,and the optimized redundant complementary dynamic Alliance coordination scheme is proposed.Simulation experiments are carried out to determine the optimal performance state for large-scale sensor cooperative applications.(2)In multi-sensor data association.Firstly,the tracking gate that most of the correlation methods don't pay much attention to is analyzed in depth.In the aspect of traditional tracking gate optimization,we try to put forward the tracking gate optimization method of genetic algorithm fusion coding,and then focus on the theoretical characteristics of traditional data association algorithm,and put forward particle swarm annealing association algorithm.Finally,the multi-sensor image information fusion is studied.The level of image information fusion,fusion technology and fusion process are discussed in detail.The discrete wavelet and color space registration methods are optimized,and good results are achieved through experiments.(3)In multi-target trajectory prediction and tracking.Firstly,the Kalman filtering theory of target state,the basic assumption of target motion state and the conventional algorithm and scheme of current target tracking technology are introduced.Then,an optimized SSUKF algorithm is proposed for the deficiency of current tracking scheme.Combined with the optimization of cooperative networking mode in Chapter 2,multitarget tracking is carried out.Matlab2019 b simulation software is used to simulate and realize multi batch target state tracking.Compared with the extended Kalman filter,the spherical distributed unscented Kalman filter has better accuracy and robustness.