Study On Space Registration Technology For Multi-sensor Information Fusion

Multi-sensor information fusion system is an important technical means which can effectively cover information domains including time, space, frequency.etc, and improve decision efficiency. It's applied widely in military and civilian applications. Space registration is the pre-processing step in information fusion. It can realize space information unification between multi-sensor, multi-platform and multi-systems which are set with different coordinate systems or different positions. It is also an indispensable step to complete high accuracy fusion detection, tracking, identification and advanced analysis. Generally space registration involves two kinds of methods: offline estimation and online estimation. For complex battlefield environment, the latter method can provide more accurate description of instantaneous position error arouse by various interference.UKF is an effective strategy for online estimation. It does not need compulsory linearization of nonlinear systems and does avoid tedious calculation of Jacobin matrix. Furthermore, its convergence speed and registration precision is higher than EKF with same calculation load. So it can achieve state estimation and nonlinear spatial location error compensation at the same time.One difficulty in missions of multi-sensor space registration is the nonlinear factor on condition that the sphere coordinate system is transformed to the Cartesian coordinates system. Using UKF, the nonlinear online position error can be estimated more effectively. Under the same perception range and same combination mode of sensors, based on prior knowledge obtained by cooperative targets, this paper proposed a method which use spatial error information of cooperative targets to improve the spatial position precision of non-cooperative targets further.Considering engineering demands, this paper analyze the performance and adaptability of proposed method mainly on the following aspects: choice of the coordinate origin, choice of coordinate axis and distance between sensors by measurement results and registration precision, etc.Based on the above research, we designed some simulation experiments. The results verified the effectiveness of the proposed method.