Study Of Feature Analysis And Target Recognition For High Resolution Range Profile

While these complex and strict modern tasks cannot be satisfied by targetdetection and tracking position of traditional radar, radar target recognitiontechnology springs up and has broad developing space in many fields.The highresolution range profiles (HRRP) can achieve target’s high resolution imaging alongthe radial distance and contain enough information of targets. And because ofobtaining easily and good real-time, HRRP is one of the important features of radartarget recognition. This paper focuses on the study of feature analysis andtarget recognition for HRRP with electromagnetic simulation technology.In radar target recognition research, it is very important to obtain effectiveradar target signature signal. This paper studies the application of electromagneticscattering simulation method for HRRP. By geometric modeling, theoreticalestimation and calculation method selecting reasonably,we can analyze theelectromagnetic scattering properties of complex targets and obtain effectivescattering field data using the electromagnetic analysis software in computersimulation environment.Then these data are transformed into HRRP. At the sametime, the simulation result for the ship is presented,which shows that the method cansimulate the features of complex targets accurately to some extent.The high resolution range profiles is sensitive to the change of the target’sattitude angle,which will cause problems to target recognition. Therefore theeffective feature extraction is particularly important.The feature extraction ofsubspace based criterion function can put forward the criterion function andestablish a subspace to realize feature extraction.This method can not only improvethe attitude angle sensitivity, but also realize dimension compression.Kernellearning based feature extractions, which is capable of implicitly mapping nonlinearfeature spaces,can hand complex and high-precision tasks. So it is widely used inthe study of pattern recognition. This paper studies the subspace method and kernellearning based feature extractions,which are applied to HRRP as target identification.A comprehensive analysis is presented after comparing results,which indicates thatthe recognition method is effective and practical by the simulation of the targets’example.