Two-dimensional matched filtering feature extraction and classification detection of active sonar echo

To reduce the clutter false alarm in active sonar echo detection, and enhance the echo detection rate while maintaining a low false alarm rate, a classification detection method based on the two-dimensional matching filter (2D-MF) is proposed. This method divides the matched signal into multiple sub-signals with appropriate pulse widths, and the sub-signals are used to respectively perform the matched filtering and extract the 2D-MF features from the active sonar received data. The extracted 2D-MF features utilize the time-frequency information and matching gain simultaneously. The convolutional neural network is then employed as the echo detector, effectively distinguishing between echo signals and clutter signals. Simulation and experimental results demonstrate that this method significantly improves the echo detection rate while maintaining a low false alarm rate in shallow-water channels. Specifically, it achieves an echo detection rate of 91.30% with a false alarm rate of 1‰ for the at-sea measured data, representing an approximate 4% improvement compared to existing methods.