Study Of Feature Detection And Classification Recognition For Aerial Target

Due to the characteristics of high maneuverability and high flexibility,aerial targets are extensively applied in modern battlefield and a variety of important tasks are undertaken by them.The research of detection and classification recognition for aerial targets not only contributes to the timely discovery of invasive targets in air defence early warning system,but also helps to obtain the specific type and the mounted weapons,providing a powerful technical support for battlefield command.With the complication of electromagnetic environment and the increasing of the target type,there are low-altitude slow-moving targets,such as hovering helicopter.In this case,the echoes of targets are easily submerged in strong noise or clutter,resulting in the poor performance of conventional detection algorithms.In order to cope with the limitation of conventional threshold detection and the desperate need of classification recognition of airplanes,the study of feature detection and classification recognition for aerial target is performed based on the analysis of modulation characteristics.Specifically speaking,the main content of this thesis is as follows:1.Based on the micro-motion model,the features corresponding to three aerial targets whose echoes are obtained by a low-resolution radar system are extracted and analyzed.Firstly,the analysis of micro-motion models corresponding to helicopter,propeller and turbojet is performed.The simulation experiments are proceeded to acquire the corresponding echoed data.By analyzing the echoes,it is found that the basic form of three targets is similar,manifesting as strong fuselage component and a certain proportion of micro component.The differences lie in their micro components.Based on the difference,the modulation characteristics are investigated and a variety of features including the waveform and time-frequency features are extracted to reflect their modulation characteristics.Finally,the micro-Doppler of helicopter with different blade shapes is introduced and particularly analyzed.The theoretical analysis and the measured data are combined to discuss the impact brought by the blade shapes,indicating its potential value.2.Considering that the conventional detection algorithms do not take account into the modulation characteristics induced by targets,the feature detection method which can be applied to low-altitude slow-moving targets,is studied.Firstly,the profound discrepancy between targets and noise or clutter is found.In accordance with this property,the feature distribution of target returns and noise is empathetically studied,evaluating the separability of corresponding features.Based on the analysis result,several features reflecting the unique modulation characteristics are selected and extracted.Then,a method of feature detection is presented.The experimental result justifies that the given method can effectively achieve the detection of low-altitude slow-moving targets.3.Based on the feature analysis of measured data,the classification and recognition of three aircrafts in low resolution radar system are studied.The analysis of three aircrafts,corresponding to the helicopter,the propeller,and the turbojet,is given.It is found that these airplanes display the evident difference in the micro-Doppler signatures.Then,some features are extracted and compared to select the distinguished ones.After that,an algorithm is proposed by using multiple features to classify the aircraft targets.After the realization of classification,the recognition of two helicopters is discussed based on the time-frequency signature analysis.The processing result of measured data shows that the proposed method can effectively classify and recognize targets.