基于多信息字典学习及稀疏表示的SAR目标识别  

SAR target recognition based on multi-information dictionary learning and sparse representation

作　　者：齐会娇 王英华[1] 丁军[1] 刘宏伟[1] 

机构地区：[1]西安电子科技大学雷达信号处理国家重点实验室,陕西西安710071

出　　处：《系统工程与电子技术》2015年第6期1280-1287,共8页Systems Engineering and Electronics

基　　金：国家自然科学基金(61201292;61322103;61372132);全国优秀博士学位论文作者专项资金(FANEDD-201156);国防预研基金;中央高校基本科研业务费专项资金资助课题

摘　　要：为了提高合成孔径雷达(synthetic aperture radar,SAR)图像中目标变体的识别性能,在鉴别字典学习及联合动态稀疏表示模型的基础上,提出一种基于多信息字典学习及稀疏表示的SAR目标识别方法。在训练阶段,采用鉴别字典学习LC-KSVD方法分别对目标图像域幅度信息及目标频域幅度信息进行字典学习。在测试阶段,结合训练阶段学到的2种信息的字典及测试目标的2种信息,采用联合动态稀疏表示模型求解2种信息下的稀疏表示系数。最后,根据2种信息下的重构误差实现对测试目标的识别。使用MSTAR数据集对算法进行验证,结果表明,新方法相对于现有的方法能够达到更好的识别性能。To improve the synthetic aperture radar （SAR）target variant recognition performance, on the basis of the discriminative dictionary learning and joint dynamic sparse representation model, a new SAR target recognition method is proposed based on the multi information dictionary learning and sparse representation. In the training stage, the discriminative dictionary learning method label consistent KSVD （LC-KSVD） is used to learn dictionaries for both the image domain amplitude information and the frequency domain amplitude information of the targets. In the test stage, based on the learned dictionaries for the two kinds of information, the test target representation coefficients for the two kinds of information are computed using the joint dynamic sparse representation model. Finally, the test target can be classified according to the representation residual for the two kinds of information. The MSTAR dataset is used to verify the effectiveness of the proposed method. Experimental results show that the proposed method has better recognition performance than some existed methods.

关 键 词：合成孔径雷达 目标识别 字典学习 联合动态稀疏表示 

分 类 号：TP95[自动化与计算机技术]