Study On Recognition Of Radar Target Using Range Profiles

The radar target recognition, which plays an important role in modern radar, has been one of the key components of the present and future defense weapon system. The range profiles of target can be obtained easily by high-range resolution radar (HRR). The range profiles contain more information for recognition. Many methods of feature extraction and classification in radar target recognition using range profiles are studied intensively in this paper. From the point of view of differential geometry, we take into account the relation of the feature extraction and classification reasonably. The issue of rejection theory and classifier combination for radar target recognition is studied too.Main contents and innovations of this dissertation are given below.1. Put forward Orthonormal Discriminant Analysis-based subspace feature extraction methods in radar target recognition. And also introduce genetic algorithm to choose the best feature for PCA and ICA. In addition, many subspace target recognition methods, obtained by combining different feature extraction methods with different classifiers, are studied.2. Put forward non-linear recognition methods in radar target recognition. Through the introduction of kernel function, PCA and LDA are easily expanded to nonlinear field, and thus KPCA and KDA are obtained. Recognition performance of KPCA, KDA and SVM in radar target recognition is studied. The non-linear recognition methods based on KPCA and KDA are more effective than linear recognition methods based on PCA and LDA.3. Study three kinds of recognition methods on the basis of Bayesian networks theory in radar target recognition. The recognition methods based on Bayesian networks theory are effective, and can meet the real-time performance requirements in radar target recognition.4. Put forward Riemannian metric-based nearest center neighbor classifier (RMNCC) on the basis of differential geometry, correlating the feature extraction based on PCA and ICA and classification in radar target recognition on the foundation of mathematics. Put forward advanced PCA and LDA subspace target recognition methods, considering the relation of data distribution, feature extraction and...