Research On Millimeter Wave Multiple Infrared Scanning Detection And Target Recognition Method

With the complication of modern warfare,the application requirements for intelligent ammunition of high-precision attack performance on the battlefield are increasing,and higher-level requirements are put forward on target detection and recognition technology.Due to harsh environments such as haze,rain,snow,and smoke can cause serious interference to target detection,resulting in incomplete acquisition of target information by weapons with a single detection technology,which is prone to miss and false detections.As to this problem,This paper adopts millimeter wave/infrared composite detection technology to give full play to its advantages of anti-jamming,anti-stealth,all-weather,and carries out in-depth research on the accurate identification of ground armored targets.Firstly,based on the principle of compound detection,the millimeter-wave multiple infrared compound push-broom model and the projectile intersection model are established,and the field of view layout of the millimeter-wave radiometer and infrared detector is designed.Based on the measured surface temperature data of Type 59 tank,the distribution of target surface temperature field was obtained by Fluent finite element analysis of tank geometric model.According to the temperature field distribution,the radiation model is calculated,and the response simulation of the 8mm millimeter wave radiometer and the 16-element infrared detector is completed.Secondly,the signal oscillation is caused by hard threshold discontinuities,as well as the signal edge ambiguity is caused by soft threshold functions are improved through designing an improved threshold function,to obtain millimeter wave and infrared signals with high signal-to-noise ratio,low mean square error,and high correlation with the original signal.Furthermore,the characteristic quantities of millimeter wave and infrared signals are counted,and an algorithm based on Adaptive Neural Fuzzy Inference System(ANFIS)is used to achieve the single-mode target recognition.In view of the problem that the traditional D-S evidence theory algorithm has evidence conflict in the fusion process,which causes the target to be misidentified or unrecognized,this paper proposes to modify and improve the evidence source twice to get the new basic probability distribution value and complete its own data fusion.Finally,in order to simulate the scanning process of the glancing target of the missile body,the experimental environment of the sliding lock in the field is built to complete the multi-angle rendezvous detection experiment of the projectile.The data output by the detection system is analyzed and processed by MATLAB.The results show that the recognition rate of ANFIS reaches 83.47%,and the recognition rate of improved DS evidence theory reaches 90.48%,which is 2.86%higher than that of traditional D-S evidence theory and about 7%higher than single-mode recognition accuracy.