Abstract: Feature extraction is a key step of underwater passive sonar target classification and recognition.A kind of tensor feature extraction method based on auditory Patterson-Holdsworth cochlear model is proposed.First,the filter impulse response of the cochlear model is regarded as the basis function of signal decomposition,and the center frequency of different channels is determined according to the nonlinear scale or conventional linear scale of the auditory model.Then,the gain and bandwidth of the corresponding channel are calculated,and the order and phase parameters of the impulse response are quantified to obtain a signal decomposition basis.And according to the principle of signal decomposition,the third-order tensor features of channel number-order-phase number are obtained.Finally,the classification and recognition of underwater passive sonar target is realized by calculating the similarity between test sample tensor feature and training sample tensor feature.The experiment of passive sonar target classification and recognition shows that the extracted tensor features have better classification and recognition performance,and the equivalent rectangular bandwidth scale of auditory model is better than the linear scale to divide the center frequency,which can improve the target indication ability of passive sonar.