High Range Resolution Profile Target Recognition Based On Compressed Sensing

Target recognition method based on one dimension range profile is one of the main research fields of modern radar target recognition. With the rising of the Compressed Sensing theory, people tried to apply it to target recognition. This paper mainly studies how to extract different features and design different classifiers based on CS theory, to finally achieve the purpose of radar HRRP target recognition. The main content is summarized as follows:1. The method of radar HRRP target recognition based on the sparse representation theory is studied. In this method the testing sample is recognized according to its sparsity on the redundant dictionary formed by the training samples. This method uses the correlation of the testing samples and the training samples directly, avoiding the problem of features selection. Simulation experiments prove that this method can achieve a good recognition effect in small attitude angle range.2. A method of radar HRRP target recognition based on the K-SVD dictionary learning method is studied. Adaptive dictionary is more consistent with the characteristics of the samples, leading to an increasing sparsity of the testing sample on the dictionary, so that can improve the effect of the recognition. Simulation experiments prove that this method can improve the recognition rate further, overcome the sensitivity of HRRP attitude angle to some extent, and achieve a good recognition effect in a large attitude angle range.3. A method of radar HRRP target recognition based on the discriminative K-SVD(D-KSVD) dictionary learning method is studied, which generates a linear classifier when learning the overcomplete dictionary, so it is simple and intuitive, avoiding the feature extraction and the classifier design. Simulation experiments show that in a small attitude angle range, this method can reach a high recognition rate.4. A method of radar HRRP target recognition based on the matching dictionary is studied. In this method we extract the relative position and strength of the scattering point on the target by calculate the sparse representation coefficients of the target HRRP on the matching dictionary, and then use the SVM classifier to identify the target. Simulation experiments prove that this method can accurately extract the information of the scattering point, and achieve a good recognition effect.