基于KPCA準(zhǔn)則的SAR目標(biāo)特征提取與識別
SAR Automatic target recognition based on KPCA criterion
-
摘要: 該文給出了一種基于 KPCA(Kernel Principal Component Analysis)和 SVM(SupportVector Machine)的合成孔徑雷達(dá)(Synthetic Aperture Radar,SAR)目標(biāo)特征提取與識別方法。該方法在非線性空間內(nèi)利用線性 PCA(Principal Component Analysis)準(zhǔn)則提取目標(biāo)特征并由 SVM分類器完成目標(biāo)識別?；诿绹鴩栏呒壯芯坑媱澥?Defense Advanced Research Project Agency,DARPA)和空軍研究室(Air Force Research Laboratory,AFRL)提供的實測 SAR地面目標(biāo)數(shù)據(jù)的實驗結(jié)果表明,該文方法不但能夠提高識別率,具有良好的推廣能力,同時還降低了對方位估計精度的要求,是一種有效的 SAR目標(biāo)特征提取與識別方法。Abstract: In this paper, SAR ATR (Synthetic Aperture Radar Automatic Target Recogni-tion) approach based on KPCA (Kernel Principal Component Analysis) is proposed. KPCA first maps the input data into some feature space using kernel functions and then performs lin-ear PCA on the mapped data. It takes the principal components in nonlinear space as sample features, then SVM classifier is used to classify targets. Experimental results with MSTAR SAR, data sets provided by the US DARPA/AFRL (Defense Advanced Research Projects Agency/Air Force Research Laboratory) show a better performance of classification and generalization.
-
B. Scholkopf, A. Smola, K. R. Mller, Nonlinear component analysis as a kernel eigenvalueproblem, Neural Computation, 1998, 10(5), 1299-1319.[2]V.N. Vapnik.[J].Statistical learning theory, ATT Research, London University.1998,:-[3]E.R. Keydel, S. W. Lee, JT. Moore, MSTAR extended operating conditions, A Tutorial, SPIE,1996, 2757(3), 228-242.[4]Qun Zhao, DongXin Xu, J. C. Principe, Pose estimation of SAR automatic target recognition,Proceedings of Image Understanding Workshop, Monterey, CA., 1998, 11,827-832.[5]T. Ross, S. Worrell, V. Velten, J. Mossing, M. Bryant, Standard SAR ATR evaluation experimentusing the MSTAR public release data set, SPIE, 1998, 3370(4), 566-573.[6]Qun Zhao, J. C. Principe, Support vector machine for SAR automatic target recognition, IEEETrans. on Aerospace and Electronic Systems, 2001, 37(2), 643-654. -
計量
- 文章訪問數(shù): 2681
- HTML全文瀏覽量: 157
- PDF下載量: 510
- 被引次數(shù): 0
下載:
