基于训练特征空间分布的雷达地面目标鉴别器设计

摘要

To identify the out-of-database targets in the process of radar ground target recognition with High Resolution Range Profile (HRRP), this paper proposes an improved radar ground target identifier based on the distribution of the space of training features. In the training phase, a K-Means clustering strategy based on the pre-process of correlation coefficient is utilized to divide the space of training dataset. Then each sub-space boundary is determined by Support Vector Domain Description (SVDD) based on the distribution of the sample space. Finally, it can decide the category of target with the sub-space boundary and the weighted K-neighbors principle. This method can work without the template of out-of-database samples, which improves the effectiveness of target identification. Due to the fact that the feature space of different targets has the characteristic of non-uniform aggregation under different attitudes, a procedure of region partition is applied to training dataset. Thus computational load is relieved with a decrease in search operation of template matching. The requirement of real-time processing can be satisfied. Finally, the experiments against both simulation and real data verify its excellent performance of identification and real-time processing capability.%该文对雷达地面目标高分辨1维距离像目标识别中的库外目标鉴别问题，提出一种基于训练特征空间分布的雷达地面目标鉴别器。在训练阶段利用基于相关系数预处理的K-Means聚类方法对库内目标样本特征空间进行区域划分，并采用基于空间分布的支撑向量域描述方法确定样本特征空间的边界与支撑向量，利用样本特征空间边界与加权 K 近邻原则对目标类别进行判决。该方法解决了库内目标与库外目标的鉴别问题，提高了目标识别系统的总体性能。针对多种不同姿态下目标特征空间非均匀聚合的特点，对训练样本特征空间进行区域划分，减小模板匹配搜索运算规模，保证目标鉴别所需的实时性工作要求。最后通过仿真和实测数据验证了该方法具备优良的鉴别性能与良好的实时处理能力。