给出了一种基于KFD和SVM的合成孔径雷达目标特征提取及识别方法。该方法首先在非线性空间内利用Fisher准则提取样本特征 ,然后由SVM分类器完成目标识别。实验结果表明该方法不但有效地提高了目标的正确识别率及运算效率 ,还大大降低了对目标方位的敏感度 ,在目标方位信息未知的情况下 ,识别率仍可达到 95 %以上

In this paper, a method of SAR ATR (Synthetic Aperture Radar Automatic Target Recognition) based on KFD (Kernel Fisher Discriminant) is described. It consideres the projections of samples onto an optimal vector derived by Fisher's discriminant in nonlinear space as sample features, then SVM classifier is used to implement target identification. Experimental results show better performance of identification and less running time. Moreover, we conclude that good experimental results can be obtained even if no target azimuth information is knawn. [

TN957.52

国家自然科学基金项目(6690200 9，60272409)，中科院自动化所模式识别国家重点实验室开放课题，中国民航总局教育研究基金