Abstract We propose in this paper a new method for targets recognition from radar images. To characterize the radar images, we adopt a statistical multivariate modeling using copula in the complex wavelet domain. For the recognition step, we investigate the weighted sparse representation-based classification (WSRC) method. To build the dictionary, the estimated copula parameters are stacked together in a matrix structure. In order to include the locality information of this dictionary for each unknown radar image to recognize, we affect weights for its atoms (columns). That is done by calculating the Kullback–Leibler divergence (KLD) between the multivariate copula parameters of training and test radar images. Finally, the unknown radar image is recognized through the SRC classifier. Several empirical results carried out on the SAR (synthetic aperture radar) and ISAR (inverse synthetic aperture radar) images demonstrate that the proposed method achieves high recognition rates and outperforms the remaining methods.

中文翻译：

---

用于雷达图像目标识别的多元Copula统计模型和加权稀疏分类

摘要 本文提出了一种从雷达图像中识别目标的新方法。为了表征雷达图像，我们采用了在复小波域中使用 copula 的统计多元建模。对于识别步骤，我们研究了基于加权稀疏表示的分类 (WSRC) 方法。为了构建字典，估计的 copula 参数以矩阵结构堆叠在一起。为了包括该字典的位置信息以供识别每个未知雷达图像，我们影响其原子（列）的权重。这是通过计算训练和测试雷达图像的多元 copula 参数之间的 Kullback-Leibler 散度 (KLD) 来完成的。最后通过SRC分类器识别未知雷达图像。