基于模糊超球面支持向量机的超宽带SAR地雷检测

机载或车载超宽带合成孔径雷达(UWB SAR)可以大区域快速探测单个地雷和雷场，是探雷的发展趋势。虚警太多是UWB SAR探雷实用化的主要问题。本文提出了模糊超球面支持向量机( FHS-SVM)地雷检测器。FHS-SVM在高维核特征空间中构造封闭的超球面区分地雷和杂波，并在学习过程中利用隶属度定量表征地雷和杂波误判风险不同及地雷埋设环境变化等因素对检测器的影响。轨道地表穿透SAR(Rail-GPSAR)系统实测数据处理结果表明，FHS-SVM比传统超球面SVM(HS-SVM)和超平面SVM(HP-SVM)具有更好的检测性能。

Ultra-wideband SAR Landmine Detection Using Fuzzy Hypersphere Support Vector Machine

Air- or vehicle-borne ultra-wideband synthetic aperture radar (UWB SAR) can be used to detect landmines or minefields over large areas quickly, wmch is the present trend in landmine detec?tion. Yet it leads to too many false alarms in practical applications. In this paper, a fuzzy hypersphere support vector machine (FHS-SVM) landmine detector was proposed. The FHS-SVM forms a hyper?sphere in the high dimensional kernel feature space to separate landmine and clutters, where the fuzzy membership is introduced in the learning procedure to quantitatively represent the effects of misclassifi- cation risk difference between landmines and clutters and landmine buried environment varying on de-tector .The FHS-SVM is validated by the rail ground penetrating SAR (Rail-GPSAR) system field da?ta. The validated result shows that the FHS-SVM outperforms the conventional kypersphere-SVM (HS-SVM) and hyperplane-SVM (HP-SVM) in landmine detection.