一种基于HMM的雷达目标识别方法

Approach Based on HMM Framework to Identify Radar Target

根据雷达目标散射点的一般模型和特定条件的简化模型 ,提出用 REL AX方法从高分辨雷达回波提取目标散射点分布的位置信息作为目标 HRRP的特征向量 ;利用 HRRP对雷达视角敏感这一特点 ,用隐马尔可夫过程表征多视角雷达回波序列 ,获得目标距离 -方位两维信息 .用若干 HMM子过程构成的模式链表征一个飞行目标的飞行姿态变化 ,从而采用基于隐马尔可夫模型的分类器实现目标类属和方位的自动识别和分类 .实测数据的计算机仿真结果表明 ,这一方法的平均识率为 99.80 %和 81.2 % .

Based on general models of the radar target scatter and predigested models under the special conditions, an approach is proposed to produce radar vectors from high resolution radar profiles, here the method of vector quantization uses RELAX algorithm. The hidden Markov model is used as the classification algorithm. HMM is used to describe multi-azimuth radar profiles to obtain the distance-azimuth information of radar target through that radar HRRP is sensitive to radar azimuth. The usultsot computer simulation shows that the identification accuracy adapted method is very high. The average identifying rates are 99.80% and 81.20% in different data.