基于Markov模型的机动目标分类方法

Maneuvering Targets Classification Method Based on Markov Model

结合弹载脉冲多普勒雷达精确提取目标速度、加速度的特点,研究了利用目标的机动特性进行目标分类的方法。首先提出对目标加速度序列进行标量量化的预处理方法,然后基于一阶马尔可夫(Markov)模型建立了用于对目标加速度序列进行识别的分类器并定义了模型距离,最后分析了共用码书所刻画的状态空间对分类性能的影响。仿真实验证明,该算法计算量小,易于实现,在较小的模型距离下,能取得较高的识别率。

Using the character of airborne pulse Doppler radar that can extract the velocity and acceleration of targets, a maneuvering targets classification method based on Markov model has been proposed. First, a preprocessor which applies scalar quantization method to acceleration series is introduced. Then, a classifier based on Markov model is designed to perform the classification according to targets' acceleration feature. A kind of model distance is defined to demonstrate the performance of proposed classifier. At last, the effect of the unique state space to classification performance is analyzed. The experiments illustrated that the results are promising under little model distance with less computational burden.