摘要
基于高分辨径向距离像HRRP(High Resolution Range Profile)的目标识别一直是雷达目标识别研究的重要方向。H RRP的目标姿态敏感性极大地影响了识别性能,尤其是全方位角目标识别的性能。本文提出一种基于混淆矩阵的分类方法,采用支持向量机(SVM)作为基本的两类分类器(Binary Classifier),使用H AC(H ierarchicalAgglom erative Clustering)构造了一个基于“错误纠正”策略的两层层次化分类器(H ierarchicalClassifier)。实验表明,在复杂度增加不大的情况下,识别性能得到了相当程度的提高。
It has been an important area on the research of radar target recognition based on HRRP(High Resolution Range Profile). HRRP is aspect-sensitive which greatly deteriorates the performance of recognition, especially that of whole-aspect-range recognition. We propose a confusion matrix based SVM classification algorithm, which constructs an‘error-correct’ hierarchical classifier using HAC (Hierarchical Agglomerative Clustering). Experiments show that the performance of recognition improves considerably at expense of limited extra complexity.
出处
《微电子学与计算机》
CSCD
北大核心
2005年第3期136-139,143,共5页
Microelectronics & Computer
