Abstract:As one of the main signal sources of underwater acoustic target recognition, the target noise signal is difficult to characterize the characteristics of the target from clearly comparing with the multi-sensor detection technology, which may lead to lower recognition rate and higher false alarm rate and seriously restricts the function of underwater acoustic detection system. In order to solve this problem, a multi-layers LSTM underwater acoustic target noise feature extraction model is established by using the long short term memory network. The information features such as time-domain envelope of target noise, Demon line spectrum and Mel frequency cepstrum coefficient are extracted, and a subset of multi-classes features is constructed. On this basis, the feature level fusion recognition and classification model based on the multi-classes features subset and the decision level fusion recognition and classification model based on the D-S evidence theory are established, and the above-mentioned models are tested by using the sample database. The difference of classification result between the multi-classes feature fusion and the single class feature recognition classification is compared, and the above model is tested and verified by using the relevant test data of port basin verification experiment. The correlation results show that the present intelligent recognition and classification method of underwater target noise based on the multi-classes feature fusion is more robust, and the recognition rate and false alarm rate of underwater target are better than those of single category feature discrimination method.

Negative Correlation Incremental Integration Classification Method for Underwater Target Recognition

Research on Underwater Small Target Detection Algorithm Based on Improved YOLOv3

Underwater Target Recognition Algorithm Under The Influence Of Non-Uniform Illumination

Selective Negative Correlation Learning Approach to Incremental Learning

Deep Learning-Based Recognition of Underwater Target

Selective negative correlation learning algorithm for incremental learning

Few-shot Underwater Acoustic Target Recognition Based on Siamese Network

Exploring Underwater Target Detection Algorithm Based on Improved SSD

Ia-Pncc: Noise Processing Method for Underwater Target Recognition Convolutional Neural Network

A New Method for Underwater Target Classification

A new approach to underwater target recognition

Underwater signal recognition based on integrating domain adaptation framework with the stochastic classifier

Negative Correlation in Incremental Learning.

Underwater Target Classification Based on Feature Fusion and Gene Encoding of CNN-BIGRU-Attention

Underwater Target Noise Recognition and Classification Technology based on Multi-Classes Feature Fusion

Underwater Target Recognition Using Artificial Fish-Swarm Algorithm

Improved Deep Learning Approach For Underwater Salient Object Detection Using Spectral Residual and Fuzzy C-Means Clustering

Research on Underwater Target Recognition Technology Based on Neural Network

Underwater Noise Target Recognition Based on Sparse Adversarial Co-Training Model with Vertical Line Array

Underwater Acoustic Target Recognition Method Based on Feature Fusion and Residual CNN

Incremental Negative Correlation Learning with Evolutionary Selection of Parameters