Joint Target Detection And Feature Correlation Analysis Method Based On Microwave And Laser Radars

The radar backscattering cross-sectional area of the stealth target is small and has low observable characteristics,but it’s difficult to be stealthy both in microwave and laser bands.The combined use of microwave and laser radars can realize the utilization of target characteristics across the spectrum,and has the potential to improve the performance of stealth targets detection.Aiming at the background requirements of long-distance stealth flying target detection in the air,this paper studies microwave and laser radars target characteristics modeling and data simulation,multi-sensor joint detection and feature correlation analysis methods to improve the performance of long-distance flying targets detection and classification.The main work and innovations of the paper are as follows:Aiming at typical air targets,the modeling and data simulation of microwave and laser radars target characteristics are realized,and the idea of joint detection of low-observable targets in the air based on microwave and laser radars is established.For airborne microwave and laser radars aerial target joint detection scenarios,the echo observation model of point target and extended target are established,and the characteristics of target and echo data are analyzed,which provides theoretical basis and data support for subsequent experimental research.Aiming at the problem of radar and Lidar multi-sensor target detection,a heterogeneous sensor cascade detection method based on generalized likelihood ratio test is proposed.Firstly,considering both the sensor’s detection efficiency and detection performance,a cascade detection strategy is proposed,which uses the coarse detection results of microwave radar to guide the fine identification of laser radar.The cascade detector model and detection performance are deduced and the simulation experiments are carried out to verify its effectiveness.The simulation results show that under the same false alarm rate,the combined detection probability of microwave and laser radars is higher than that of single microwave radar detection,and the calculation amount of joint detection is far less than that of single laser radar detection.When the false alarm rate is equal to 10^-5 and the detection probability is equal to 0.8,the detection distance of cascade detection is improved by about 14%compared with single microwave radar detection for the parameter conditions in this paper.On the basis of that,a microwave and laser radars experimental system is constructed,and the cascade detection method is verified by measured data.The results show that under the same detection constraints,the detection probability of joint detection method is higher than that of single microwave radar detection.For microwave and laser radars’observation data,an improved correlation feature learning and extraction model based on Siamese neural network is proposed,and the simulation results show that the target classification accuracy is improved through correlation feature fusion.Firstly,an correlation feature learning model which is named as Siamese P3SNet is proposed,the model based on Siamese neural network model,and is embedded in the feature decomposition auto-encoder structure of the P3SNet cross-modal retrieval model,the multi-sensor correlation features are extracted through data learning.Then,the correlation features were analyzed by mutual information estimation and t-SNE feature dimensionality reduction visualization,and the representation ability of correlation features to the observation data is shown from the visual level.Finally,the SVM classification results show that correlation features extracted by the Siamese P3SNet model can effectively improve the classification accuracy by 4.58%than that of simple feature fusion classification.