针对战场声信号复杂多变的特点,提出了一种基于小波包特征参数(WPFC)和隐马尔科夫模型(HMM)相结合的战场声目标识别方法。该方法利用小波包对信号高、低频段能进行精细划分,从而得到更能反映战场声信号特征的小波包特征参数;并利用HMM具有很强的表征时变信号能力的优点,将HMM作为训练识别模型。仿真结果表明了此方法的准确性和可行性。

Since battlefield acoustic signal is complicated and fickle, a new battlefield acoustic target identification method was proposed based on wavelet packet feature parameters (WPFC) and hidden Markov model (HMM). The method uses wavelet packet to subtly divide signal's high and low frequency bands so as to obtain wavelet packet coefficients that reflect the battlefield acoustic signal better, by making use of the merits of strong ability of time-varying signals, the hidden Markov models were established as training and recognition models. The simulation results show accuracy and feasibility of the method.