基于Laplacian正则化最小二乘的半监督SAR目标识别

提出了一种基于核主成分分析(kernel principal component analysis,简称KPCA)和拉普拉斯正则化最小二乘(Laplacian regularized least squares,简称LapRLS)的合成孔径雷达(synthetic aperture radar,简称SAR)目标识别方法.KPCA特征提取方法不仅能够提取目标主要特征,而且有效地降低了特征维数.Laplacian正则化最小二乘分类是一种半监督学习方法,将训练集样本作为有标识样本,测试集样本作为无标识样本,在学习过程中将测试集样本包含进来以获得更高的识别率.在MSTAR实测SAR地面目标数据上进行实验,结果表明,该方法具有较高的识别率,并对目标角度间隔具有鲁棒性.与模板匹配法、支撑矢量机以及正则化最小二乘监督学习方法相比,具有更高的SAR目标识别正确率.此外,还通过实验分析了不同情况下有标识样本数目对目标识别性能的影响.

Semi-Supervised SAR Target Recognition Based on Laplacian Regularized Least Squares Classification

A Synthetic Aperture Radar (SAR) target recognition approach based on KPCA (kernel principal component analysis) and Laplacian regularized least squares classification is proposed. KPCA feature extraction method can not only extract the main characteristics of target, but also reduce the input dimension effectively. Laplacian regularized least squares classification is a semi-supervised learning method. In the target recognition process, training set is treated as labeled samples and test set as unlabeled samples. Since the test samples are considered in the learning process, high recognition accuracy is obtained. Experimental results on MSTAR (moving and stationary target acquisition and recognition) SAR datasets show its good performance and robustness to azimuth interval. Compared with template matching, support vector machine and regularized least squares learning method, the proposed method gets more SAR target recognition accuracy. In addition, the effect of the number of labeled points on target identification performance is analyzed at different conditions.