Abstract: Objectives To address the limitations of single-sensor perception in complex water surface environments, including unstable detection and tracking drift, a vessel detection and tracking method based on feature-level fusion and AIS-enhanced tracking is proposed. By effectively integrating multi-source heterogeneous information from marine radar, visible-light cameras, and the Automatic Identification System (AIS), the proposed method constructs a highly reliable vessel perception framework. Methods First, to mitigate missed and false detections caused by the small relative size and indistinct features of radar targets, this study designs VFFRadar-Net, a radar object detection network with visible features fusion. By leveraging cross-modal feature-level fusion, the proposed network enhances vessels detection performance. Secondly, a deformable cross-attention is introduced to facilitate feature interaction and alignment across modalities, thereby improving radar target representation. Third, a scale-adaptive loss function is incorporated into the network to alleviate the sensitivity of small-target IoU loss to positional deviations. Finally, this study develops AF-SORT(Improved SORT with AIS Fusion), a simple online and realtime tracking enhanced by AIS. By employing a multi-feature trajectory similarity measurement approach, AE-SORT achieves robust association between AIS and radar trajectories. Furthermore, AIS data helps mitigate error accumulation in tracking caused by measurement losses. Results Experimental results show that the proposed method achieves an average detection precision of 91.9% for ship targets, with tracking metrics MOTA(Multiple object tracking accuracy) and IDF1(Identity F1 score) reaching 85.2% and 76.2%, respectively.Conclusions The proposed method effectively mitigates missed detections, false alarms, and tracking drift in ship perception, significantly enhancing the perceptual capability for surface ship targets and providing reliable environmental awareness support for intelligent maritime supervision and autonomous navigation systems.