ISAR Imaging Simulation Of Space Targets And Target Recognition Based On ISAR Images

With the new requirement for radar to get more information from the battlefield, radar automatic target recognition (ATR) as a new research area appeared in the 1960s. The development of high range resolution radar has given strong support to radar ATR. And space target recognition technology has been paid more and more attention with the increasing development of space technology and electronic technology. ISAR image reflects the target's fine structure which is conducive to feature extraction and target classification. So the novel feature extraction and target recognition algorithms are studied based on both ISAR image simulation of space targets and real ISAR image data of aircrafts. The main contents are organized as follows:Firstly, space target's imaging model is built based on ISAR imaging principle. Three- Dimensional (3D) scattering point model of space target is built. And simulation of ISAR imaging is carried out using Rang-Doppler algorithm. The target recognition database hereby has been constructed.Secondly, feature extraction is studied based on ISAR images. Principle of image wavelet decomposition is applied to extract the singular value features from four sub-images. And a new feature extraction method based on feature integration of wavelet coefficients is presented. Then the principle component analysis combined with wavelet low-frequency band is proposed on the basis of studying the classic principle component analysis. Last the effectiveness of wavelet modulus maximum moment has been studied, so has the combined feature vector which contains moment invariant and the sizes of the targets features etc.Thirdly, the nearest-neighbour fuzzy classifier (NNFC) principle is introduced. And the NNFC is used to classify ISAR object images. Experiments with real ISAR image data of aircraft and ISAR image simulation of space targets show that the NNFC can effectively perform ISAR target recognition of multiple features combination.Lastly, the support vector machine (SVM) principle is introduced. The work of this dissertation is focused on category performance of Polynomial-SVM and RBF-SVM. Using of the feature extraction method above, experimental results with ISAR real data are analyzed in detail. The different feature extraction method as how to affect the recognition rate has been studied in depth. Then the performance of the classification of the various portfolio algorithms is compared. The optimum algorithms are used to identify space targets. Experiments with ISAR simulation data achieve good recognition effect. Conclusions are drawn in the end.