Research On Anti-Jamming Detection Algorithm For Infrared Small-Target

Reliable detection of small targets in complex confrontation environments is a guarantee of the performance of the detection and monitoring system in the battlefield.At present,the research on anti-jamming detection algorithm for moving small targets is a research hotspot in the field of navigation and guidance,which has important significance for improving China's military defense and strike capability.This thesis is divided into three parts,the first part proposed an algorithm for inversion of infrared small target radiation characteristics without a standard radiometric calibration.The surface features with stable radiant intensity in the experimental scene were selected as calibration objects,such as deserts.The accuracy analysis results show that the proposed calculation model is reasonable and the inversion target radiation intensity can be used for false alarm suppression.In the second part,the difference of motion mode between target and interference is the basis of its identification.This paper introduces the characteristics of the graph,which mines the characteristic changes of nodes at different moments in the graph.The analysis shows that the eigenvector centrality of the time-varying graph can effectively express the difference of the target/interference motion pattern,which can be used as the classification feature for target and interference recognition,and the recognition accuracy reaches 96%.In the third part,the difference of motion pattern between the target and interference can be regarded as a behavior pattern,therefore,the behavior recognition method can be used to discriminate the target and the interference.The paper combines detection network(Faster-RCNN),association algorithm(Deep-Sort),and 3D convolutional network(I3D)to perform target and interference behavior recognition.Interactions with surrounding objects and other objects contain important information towards understanding the actions.The attention network was proposed,and embedding it in the middle of the I3 D network.The anti-jamming detection network can achieve an average recognition rate of 98.7% and has an effective anti-jamming ability.