基于线性预测倒谱的被动声纳目标特征提取技术

从声纳员的角度出发，被动声纳目标可以被看作为一个发声体，利用线性预测倒谱从声纳目标噪声中分离出目标作为发声体的冲激响应在倒谱域中的表示，提取一组识别特征，设计神经网络分类器，对三类目标进行分类。实测数据验证了基于线性预测倒谱的被动声纳目标特征提取方法是可行的。

Technique of passive sonar target feature extraction based on LPC cepstrum

To a sonar operator, a passive sonar target can be taken as a sounder, and the impulse response of the sounder in cepstrum-domain can be drawn from the target radiated-noise by using the LPC cepstrum. LPC cepstrum of the impulse response. A and the classification experiment for three results of experiment show that the feature is useful. A set of features is extracted based on the neural network target classifier is designed, different classes of target performed. The extraction method based on LPC cepstrum is useful.