一种基于多特征提取的雷达目标识别方法研究

A Radar Target Recognition Approach Based on Multi-Feature Extraction

直接利用雷达高分辨一维距离像或是提取一维像单一特征进行目标识别难以取得理想效果。为了更好地反映雷达目标本身的物理特性,提高雷达目标识别率,对雷达目标高分辨一维距离像的频谱幅度、能量聚集区长度、散射中心数目和中心矩特征进行提取并构成多特征向量,描述雷达目标高分辨一维距离像频域、能量等物理特性,采用主成分分析方法进行特征维数压缩,基于支持向量机的方法进行分类识别,从而提出一种基于多特征提取的雷达目标识别方法。实验结果表明,采用该多特征提取方法进行雷达目标识别有助于提高识别率和减少分类时间。

The direct use of high resolution radar range profile or single feature extracted from a one-dimensional image is difficult to achieve a desired result for target recognition. In Order to reflect the radar target characteristics of the physical structure better and increase radar target recognition rate, the spectrum range, energy accumulation zone length, the number of scattering centers and central moment features of High Range Resolution Profile （HRRP） were extracted to constitute a multi-feature vector. This multi-feature vector could describe the energy and other physical characteristics of high range resolution profile. The principal component analysis was used for feature dimensionality reduction, and Support Vector Machine （SVM） was used for classification and recognition. Thus an approach was proposed for radar target recognition based on multi-feature extraction. Experimental results show that the multi-feature extraction is helpful for improving the recognition rate and reducing the time for classification.