Acoustic Helicopter Recognition via Convolutional Neural Network

To improve the performance of recognition of acoustic target, a novel deep learning framework is proposed to extract features and recognize helicopters. In this framework, convolutional neural network (CNN) is employed to optimize the representation of the acoustic characteristics of helicopters and implement helicopter type recognition. The optimized features extracted by CNN is designed to overcome the lack of retaining intrinsic characteristics of helicopter in classical}{artificial features. Furthermore, pretreatment based on deep learning is studied for removing background noise. The proposed method was tested using the real acoustic helicopter signals collected in the field experiments. The results indicated that the proposed method significantly improves the recognition accuracy compared with conventional acoustic helicopter recognition methods. And the pretreatment method is effective for improving recognition rate when helicopter signals are interfered.