节点文献
基于核函数的雷达目标一维距离像识别研究
【作者】 张琴；
【导师】 周代英；
【作者基本信息】 电子科技大学 ， 信息获取与探测技术， 2008， 硕士
【摘要】 现代高分辨雷达的兴起为目标识别提供了新的途径。高分辨一维距离像反映了目标沿雷达径向的几何结构分布,包含了许多对目标识别有用的信息,其成像过程中避免了复杂的运动补偿问题,比二维或三维成像获取更容易。因此,近年来高分辨雷达目标一维距离像识别受到了广泛关注。本文针对一维距离像,对多种基于核函数的雷达目标识别方法进行了研究。其主要内容如下:1.介绍用于微波成像的目标散射中心模型。2.传统零空间方法充分利用了类内散布矩阵零空间中对分类的有用信息,但是当样本数很大时,传统零空间方法失效。本文研究了一种基于核函数的改进零空间方法。该方法较传统零空间方法简单,仅需分析一个特征值,且适用于大样本数的问题。3.研究了一种核主成分分析方法和线性判别分析相结合的识别方法(KPCA+LDA)。该方法结合了主成份分析方法与线性判别分析的优点,能够在保留类别主要特征的同时,最大化分类间隔。因此,具有较单一方法更好的识别性能。4.提出了一种基于核支持向量最优变换矩阵的一维距离像识别方法。该方法利用支持向量构建类间散布矩阵和类内散布矩阵,结合零空间特性得到最优变换矩阵,以提取目标特征。该方法能够提高一维距离像的识别率。本文中的方法对多组仿真目标一维距离像数据和实测飞机一维距离像数据的识别试验所验证。
【Abstract】 The increasing availability of high resolution range (HRR) radars provides a new way for radar target recognition. High resolution range profile (HRRP) shows the target’s scatters distribution along the radar line-of-sight, which contains potentially discriminative information about the target geometry. Furthermore, the HRRP can be easily acquire and also avoids the complex motion compensation processing, relative to two-dimensional or three-dimensional imagery. Therefore, HRR radar target recognition has received extensive attention from the radar technique community in recent years.Several methods based-on kernel function of radar target recognition are intensively and extensively studied in this paper. The main contents are as follows:1. The scatter-center model is discussed. The Six kinds of simulated point targets are designed and the range profiles at aspect angle are computed.2. The null-based LDA take full advantage of the null space while the other methods remove the null space, but it can not to resolve the large sample size problem. A new null space method is discuss, which is simpler than all other null space approaches and is also applicable to the large sample size problem.3. A two-phase KFD method: kernel principal component analysis(KPCA) plus Fisher linear discriminant analysis (LDA). This framework provides novel insights into the nature of KFD, it can make full use of two kinds of discriminant information, regular and irregular, and it has a better classification performance than single method.4. Proposes a novel approach, which constructs a between-class scatter matrix and a within-class scatter by use of Kernel Support Vectors (KVs). In additional, the null-space Fisher method is exploited to calculate the optimal transform matrix. This method achieves good recognition performance for HRRP.All of these methods are proved by experiments on simulated data and real data of planes. These methods can be applied to automatic and real time recognition systems.