针对SAR图像中舰船目标识别的问题，提出了基于核主成分分析（Kernel Principal Component Analysis，KPCA）和核Fisher判别分析（Kernel Fisher Discriminate Analysis，KFDA）相结合的舰船目标识别算法。用核主成分分析的方法对实测的SAR舰船目标数据进行特征降维，再结合核Fisher判别分析法对降维后的样本数据进行多类别分类。将该方法用于对实测的四类舰船目标进行识别，平均识别率可达91.25%。实验结果表明，核主成分分析与核Fisher判别分析相结合的方法可提取目标的有效特征，在较低特征维数情况下获得较高的目标正确识别率。
Ship targets recognition algorithm combining Kernel Principal Component Analysis (KPCA) and Kernel Fisher Discriminate Analysis (KFDA) was proposed to deal with the problem of ship targets recognition in SAR images. Firstly, KPCA algorithm was used to transform the sample data of high dimension space to low dimension space to reduce the dimension. Then, the processed samples were recognized according to KFDA algorithm. The method is applied for recognizing fourth-class ship targets and the average recognition arrives at 91.25%. The result showed that the combination of KPCA and KFDA can effectively eliminate the interaction between sample variable indicators. It is an effective method for SAR images feature extraction and target recognition.
2017,39(7): 149-152 收稿日期：2016-09-03
DOI：10.3404/j.issn.1672-7649.2017.07.032
分类号：TP391
基金项目：国家自然科学基金资助项目（61179016）
作者简介：刘磊(1987-),男,博士研究生,研究方向为SAR图像舰船目标识别
参考文献：
[1] DUDGEON D E, LACOSS R T. An overview of automatic target recognition[J]. The Lincoln Laboratory Journal, 1993, 6(1): 3–10.
[2] OUEHI K, TAMAKI S. Ship detection based on coherence images derived from cross correlation of multilook SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing letters, 2004, 1(3): 184–187.
[3] ROSS T. D, et al. SAR ATR: so what’s the problem An MSTAR perspective. Proc. SPIE Conf. on SAR, 1999, 3721: 662–672.
[4] 张晰, 张杰, 纪永刚. 基于结构特征的SAR船只类型识别能力分析[J]. 海洋学报, 2010, 32(1): 146–152.ZHANG X, ZHANG J, JI Y G. The capability analysis of ship classification by structure feature using SAR images[J]. Acta Oceanologica Sinica, 2010, 32(1): 146–152.
[5] SCHOLKOPF B, SMOLA A, MULLER K. Nonlinear component analysis as a kernel eigenvalue problem[J]. Neural Computation, 1998, 10: 1299–1319.
[6] 韩萍, 吴仁彪, 王兆华, 等. 基于KPCA准则的SAR目标特征提取与识别[J]. 电子与信息学报, 2003, 25(10): 1297–1301.HAN P, WU R B, WANG Z H, et al. SAR automatic target recognition based on KPCA criterion[J]. Journal of Electronics & Information Technology, 2003, 25(10): 1297–1301.
[7] MULLER K, MIKA S, RATSCH G. An introduction to kernel based learning algorithms[C]// IEEE Tran. on Neural Networks, 2001(12): 181–201.
[8] 徐正光, 王淑盛, 刘冀伟,等. 基于主成分分析的核Fisher判别方法在油水识别中的应用[J]. 北京科技大学学报, 2005, 27(1): 126–128.XU Z G, WANG S S, LIU J W, et al. Application of kernel Fisher method based on primary factor analysis to recognition problem between oil layer and water layer[J]. Journal of University of Science and Technology Beijing, 2005, 27(1): 126–128.
[9] HUANG S J, YE J Y, Wang T Q. Extracting Refined Low-Rank Features of Robust PCA for Human Action Recognition[J]. Arabian Journal for Science and Engineering, 2015, 40(5): 1427–1441.
[10] TAE H O, J C BAZIN. Partial sum minimization of singular values in robust PCA: algorithm and applications[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(4).
