Technology Research On Attributes Information Fusion Of Radar/Infrared Composite Guidance

Multi-mode composite guidance weapons,known for their high tracking accuracy,precise target striking and strong anti-jamming performance,are the focus of research on precisionguided weapons today.The radar sensor has a long observation range and is suitable for all-weather operation,but the active mode is highly susceptible to interference such as electronic detection and electronic countermeasures.The infrared sensor has high observation accuracy and is not easily detected,but its observation range is close and cannot be preempted.The combination of millimetre-wave radar/infrared thermal imaging sensors will enable a dual-mode composite guidance system to combine the detection advantages of the respective single-mode sensors and overcome the limitations and drawbacks of the respective sensors in single-mode conditions.Considering that the radar sensor data and the infrared sensor data are heterogeneous sources of information,information fusion techniques are required to integrate the data.The fusion of information in the feature layer is able to fuse the heterogeneous features of the radar and infrared sensors,eliminating useless features and redundant features,thus reducing the spatial dimension of the features and improving the recognition and classification of the composite guidance information fusion system.At present,information fusion technology has been studied and applied in some depth.However,in the engineering application of multi-mode composite guidance seeker,there are still many difficulties in the key technology of information fusion based on the feature layer in the airspace-oriented context.In this paper,the basic principles of information fusion and existing algorithms are applied to carry out further research on radar/infrared composite guidance attributes information fusion technology based on the application context of the ballistic platform.The specific research contents mainly include the following four aspects:(1)Research on the technical scheme of radar/infrared attributes information fusion.First,some basic theories of information fusion technology are briefly introduced,including the technical principle of information fusion,the multi-source information fusion model and its classification.Then,the attribute-level information fusion is introduced in detail,and the corresponding information fusion scheme is given.(2)Research on the simulation of sample data of typical air non-cooperative targets.In order to obtain the sample data from the radar sensor,the radar echo data of the target is simulated by the radar HRRP simulation method based on FEKO,is processed to obtain the corresponding HRRP data.In order to obtain the sample data of the infrared sensor,the infrared image in the same state as the radar sensor is obtained through the infrared image simulation method based on thermal analysis.By comparing with the field experimental data,it can be proved that the sample data obtained by the simulation has high authenticity,which can provide support for the follow-up research.(3)Research on feature extraction of radar and infrared data and feature selection of sensor feature sets.First,the corresponding features are extracted from radar data and infrared data,and feature selection algorithm is used to screen out feature subsets with high separability and low redundancy.Then,the algorithm used is experimentally validated in terms of both average recognition accuracy and feature space evaluation.Compared to using only a single evaluation criterion,the subset of features selected by the algorithm in this paper retains a high degree of differentiability in low dimensions,thus significantly improving the target recognition performance of the composite guidance information fusion system while ensuring real-time performance.(4)Research on feature fusion and target identification of heterogeneous information from radar sensors and infrared sensors.At the feature level,the selected radar-related feature subset and infrared-related feature subset are fused,and the feature information from the two sensors is projected and transformed into the mapping space,which can make samples have good target classification in the mapping space.After completing the offline stage of feature fusion,the relevant identification parameters required by the target classifier of the information fusion system can finally be obtained,and these parameters can be used for algorithm verification in the online stage.The experimental results show that the algorithm in this paper can obtain a higher probability of target recognition for the information fusion system.