一种融合多极化特征的雷达目标识别方法

针对高分辨率一维距离像(HRRP)多极化特征信息融合目标识别带来的数据量剧增问题,提出一种基于低维平移不变特征向量和多分类器动态组合的识别方法。该方法首先提取单极化HRRP序列的3种一维特征组成平移不变的特征向量,然后通过动态组合的方法生成总分类器组合进行分类,最后采用加权投票算法融合4种单极化HRRP的分类结果。实验结果显示,该方法在缩减数据规模的同时,有效利用极化信息,得到了较高的分类正确率。

Approach of Radar Target Recognition Based on Multiple Polarization Features Fusion

Aiming at the problem of a dramatic increase in data of multi-polarized high resolution range profile (HRRP) target recognition, a recognition algorithm based on low-dimension time-shift invariant feature vectors and dynamic combination of multiple classifiers was proposed. In this algorithm, firstly three on}dimension features of single-polarized HRRP sequence arc extracted to form the timcshift invariant feature vectors, and then a general classifier combination by dynamic ensemble of multiple classifiers is achieved, which is used to classify. Finally, the classification results of four singlcpolarized HRRPs arc assembled by weighted voting method. hhc result of experiment indicates that the algorithm not only reduces the size of the data, but also uses polarization information effectively to acquire higher correct recognition rate.