Radar Maritime Target Detection Method Based on Decision Fusion and Attention Mechanism

This paper presents a decision fusion model based on two-channel convolutional neural network (DF-TCNN) as a way to solve the problems of insufficient feature representation, low performance detection, and poor fault tolerance in radar target detection based on deep learning. In the pro-posed model, a mean strategy is incorporated on each branch’s predictions, and a decision fusion algorithm is applied to refine the classification results. Moreover, a dual-channel network structure with an attention mechanism is embedded for feature enhancement and learning adaptation. Verification of the radar data shows that the method offers a high fault tolerance rate and strong anti-interference ability, which can significantly improve radar target detection in the background of complex sea clutter.